Sanjay E. Sarma

Abstract Objective. Tacit or implicit knowledge refers to know-how that experts possess but often cannot articulate, codify, or explicitly transfer to others. This can present a significant challenge for learning, skill acquisition, and knowledge transfer across various domains, including those that rely on apprenticeships, craftsmanship, sports, and medical imaging diagnosis. This study explores whether expert tacit knowledge can be accessed and leveraged using an electroencephalography (EEG) and gaze-informed biofeedback interface to enhance expertise transfer and training. Approach. We designed an image classification task where novices were trained until they implicitly learned to classify images correctly, despite being unaware of which image regions or features guided their decisions. The task involved images with a hidden spatial asymmetry that even trained participants did not explicitly recognize. Using combined eye-tracking and EEG measures, we tracked both overt and covert visual attention to determine whether individuals unconsciously internalized this asymmetry during learning. We then investigated whether providing explicit gaze-informed feedback on their own implicit attention biases could further improve task performance of trained participants. Main results. Our findings reveal that as participants became trained, their attention patterns—both overt and covert—consistently reflected an unconscious awareness of image asymmetry, with attention biased toward task-relevant image regions. Moreover, trained individuals who received explicit feedback derived from their own gaze behavior showed additional improvements in classification performance compared to an equally trained control group. Significance. These results open the door to novel uses of biofeedback interfaces to facilitate new forms of expertise transfer, training, and collective intelligence. By extracting and conveying tacit expert knowledge—ordinarily difficult to externalize—our interface enables its transmission to novices, trained individuals, or even machine learning systems. We refer to this process as cognitive reinforcement.

CarTraC is a post-crash evidence-sharing proof-of-concept system for hit-and-run incidents, to cooperatively exchange crash data through direct V2V communication. The proposed architecture focuses on an overlooked aspect of V2V systems, post-crash evidence sharing, while prior research focuses on pre-crash safety. CarTraC detects collisions by monitoring motion data and upon detection automatically broadcasts encrypted and signed messages to nearby vehicles. These messages contain crash-related and vehicle-specific information retrievable by trusted authorities. The system operates without cellular networks or roadside infrastructure and supports two modes: Self-CarTraC, for direct data exchange, and Cooperative-CarTraC, where uninvolved vehicles assist in rebroadcasting. The security properties of the CarTraC protocol were formally verified using ProVerif. A proof-of-concept prototype was implemented on a custom ESP32-based embedded platform equipped with inertial sensors and a 2.4 GHz transceiver. Experimental validation was conducted using two stationary vehicles in an outdoor semi-urban environment, demonstrating reliable post-crash data exchange without external connectivity. The implementation achieved delivery rates of up to 100% at 15 m and 70% at 85 m, with end-to-end latencies below 34 ms.

Autonomous Vehicles (AVs) rely on sensor-based perception systems and high-definition maps for navigation. How-ever, their performance may degrade in challenging conditions such as poor visibility, unpredictable traffic conditions, or GNSS-denied environments like urban canyons or temporary construction zones. To address these limitations, we introduce Time-Logic-Map (TLM), a spatial messaging language that enables the road infrastructure to broadcast structured, machine-readable messages to supplement AV perception and support decision-making. This approach reduces the dependence on onboard sensors and describes road logic through machine-oriented language. TLM organizes road information into three layers: map, defining road geometry and a local 3D Cartesian coordinate system; logic, encoding the structural layout of roads and the precedence rules governing vehicle movements; and time, broadcasting real-time information like traffic signal phases. We describe modular design using multiple practical examples in standard and complex intersections, as well as in road construction zones.

Nowadays, Radio Frequency IDentification (RFID) systems are extensively used in many applications, such as warehouse management, inventory tracking, sensing or localization. The usage of passive RFID tags allows for low maintenance, thus reducing complexity and cost. However, in scenarios with a large population of tags, the process of discovering tags is time-consuming and may exceed the time requirements of certain applications. Rapidly anticipating the number of tags in advance, namely RFID counting, would help at optimizing the system efficiency. This work proposes an alternative conceptualization of the counting process, embedded within a novel framework that redefines collisions as a usable resource, preserving backward compatibility and contributing to lower the system latency up to 97% with a reliability of minimum 95%, depending on the scenario. Extensive system-level simulations and experimental results demonstrate that our proposed solution successfully detects an entire population of tags within a single round, offering significant advantages for various applications, including a subsequent identification phase.

We demonstrate how the RSSI and phase data from ambient RFID tags for a sleeping person can be used to monitor 4 different modalities. Prone, supine and side sleeping positions are detected with 100% accuracy when the aforementioned parameters are used as features for posture detection using machine learning. The ability to detect incontinence for fluid volumes of 10-200 ml is demonstrated. Moreover, the hydration status of a person can also be detected for a range of urine salt concentration of 20 - 280 mmol/l. We also demonstrate the ability to detect breath rate (correct to ±1 breaths/min) across 4 different individuals, resting HR of 68 beats per min (bpm) across 2 individuals and high heart rates of up to 160 bpm across one individual. We quantify the utility of the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">tag select</i> command to improve the sampling rate from targeted RFID tags to 112.63 Hz even when in the presence of a large co-located population of tags and discuss the RSSI and phase data extraction and processing in detail. Limitations of our current work and future directions are also discussed.

Collisions caused by simultaneous tag responses are a fundamental challenge in RFID systems, limiting their throughput and scalability. Existing solutions often rely on complex signal processing or hardware modifications, reducing practicality. This paper presents a novel collision resolution algorithm, fully compatible with the EPC Gen2 standard, that decodes individual tag responses by analyzing I/Q constellation patterns formed during collisions. The method introduces a new constellation cluster labeling strategy inspired by geometric alignment from computer vision, which uses analytical geometry and pattern matching techniques to efficiently resolve tag states without requiring retransmissions, channel estimation nor hardware changes. The algorithm reliably resolves up to four colliding tags and achieves up to a 33% relative gain in time-normalized throughput over Framed Slotted ALOHA (FSA). To support next-generation protocol enhancements, we also propose a multi-tag acknowledgment configuration, where the algorithm achieves up to a 121% relative gain, with peak performance at a slot-to-tag ratio of 0.5. Moreover, the algorithm achieves a 25× speedup in decoding time compared to the latest state-of-the-art method, significantly enhancing its practicality for real-time deployment. These results demonstrate the method’s effectiveness across both current and next-generation RFID systems.

A critical challenge for physical retail stores is the frequent occurrence of out-of-stock (OOS) situations, which adversely affect both sales performance and the reputation of retail establishments. In this article, we introduce an innovative methodology that exploits the multipath effect—typically considered a drawback in wireless communication—to estimate shelf availability without requiring item-level tagging and neatly arranged product placement, making it well-suited for cluttered and unstructured shelving conditions. By shifting the focus from tagging individual items to tagging the environment, this approach promises a substantial reduction in the overall costs associated with radio frequency identification (RFID) hardware. The proposed system achieves a fill level prediction accuracy of up to 97.2% for metallic, plastic, and glass packaging. Moreover, our method can efficiently differentiate three products with 90.7% accuracy while addressing scalability and privacy concerns commonly associated with existing technologies.

Secure and reliable communications are crucial for Intelligent Transportation Systems (ITSs), where Vehicle-to-Infrastructure (V2I) communication plays a key role in enabling mobility-enhancing and safety-critical services. Current V2I authentication relies on credential-based methods over wireless Non-Line-of-Sight (NLOS) channels, leaving them exposed to remote impersonation and proximity attacks. To mitigate these risks, we propose a unified Multi-Channel, Multi-Factor Authentication (MFA) scheme that combines NLOS cryptographic credentials with a Line-of-Sight (LOS) visual channel. Our approach leverages a challenge-response security paradigm: the infrastructure issues “challenges” and the vehicle's headlights respond by flashing a structured sequence containing encoded security data. Deep learning models on the infrastructure side then decode the embedded information to authenticate the vehicle. Real-world experimental evaluations demonstrate high test accuracy, reaching an average of 95% and 96.6%, respectively, under various lighting, weather, speed, and distance conditions. Additionally, we conducted extensive experiments on three state-of-the-art deep learning models, including detailed ablation studies for decoding flashing sequence. Our results indicate that the optimal architecture employs a dual-channel design, enabling simultaneous decoding of the flashing sequence and extraction of vehicle spatial and locational features for robust authentication.

In dense RFID environments, simultaneous responses from multiple tags during reader identification attempts result in signal collisions that compromise successful tag detection. These collisions, predominantly arising from the mismatch between allocated time slots and actual tag population, significantly degrade identification efficiency. The EPC Gen2 protocol employs Dynamic Frame Slotted Aloha (DFSA) to dynamically adjust frame sizes across inventory rounds, yet existing implementations lack precise tag quantity estimation during collision events. This paper presents an AI-enhanced collision resolution framework where a ResNet-based classifier analyzes collided RN16 signals to estimate concurrent responders (up to <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{1 0}$</tex> tags). By integrating this real-time collision cardinality prediction with DFSA's slot adaptation mechanism, our approach achieves up to 93.3 % time saving than conventional implementations through optimized frame size adjustments. The proposed methodology demonstrates particular efficacy in dynamic and dense deployments (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$&gt;100$</tex> tags), establishing a machine learning paradigm for physical-layer signal processing in RFID anticollision protocols.

3-Dimensional Embodied Reference Understanding (3D-ERU) combines a language description and an accompanying pointing gesture to identify the most relevant target object in a 3D scene. Although prior work has explored pure language-based 3D grounding, there has been limited exploration of 3D-ERU, which also incorporates human pointing gestures. To address this gap, we introduce a data augmentation framework– Imputer, and use it to curate a new benchmark dataset– ImputeRefer for 3D-ERU, by incorporating human pointing gestures into existing 3D scene datasets that only contain language instructions. We also propose Ges3ViG, a novel model for 3D-ERU that achieves ~30% improvement in accuracy as compared to other 3D-ERU models and ~9% compared to other purely language-based 3D grounding models. Our code and dataset are available at https://github.com/AtharvMane/Ges3ViG.

Fast and efficient RFID tag counting is an essential requirement for a lot of modern applications, particularly in high-density environments. This paper proposes an innovative algorithm designed to accelerate counting while maintaining compatibility with the EPCglobal Class 1 Generation 2 (C1G2) protocol. Relying on a single RN16 answer from each tag to validate its presence, the method speeds up counting by eliminating a full reader-tag communication, reducing processing time. The proposed method was validated using a software-defined radio (SDR) platform. Compared to the standardized protocol, our experiments demonstrated that up to four tags can be counted in the time required to identify a single tag, achieving approximately 77 % time savings across various data rates. This paper shows that the algorithm provides a feasible and scalable solution for RFID tag counting, incorporating collision information into its design to greatly improve counting speed, even in crowded environments.

3-Dimensional Embodied Reference Understanding (3D-ERU) combines a language description and an accompanying pointing gesture to identify the most relevant target object in a 3D scene. Although prior work has explored pure language-based 3D grounding, there has been limited exploration of 3D-ERU, which also incorporates human pointing gestures. To address this gap, we introduce a data augmentation framework-Imputer, and use it to curate a new benchmark dataset-ImputeRefer for 3D-ERU, by incorporating human pointing gestures into existing 3D scene datasets that only contain language instructions. We also propose Ges3ViG, a novel model for 3D-ERU that achieves ~30% improvement in accuracy as compared to other 3D-ERU models and ~9% compared to other purely language-based 3D grounding models. Our code and dataset are available at https://github.com/AtharvMane/Ges3ViG.

Vaccines are a critical public health tool to prevent deaths and hospitalizations but can be difficult to produce and distribute in sufficient quantity. Freeze-drying makes vaccines thermally stable for easier distribution but visually opaque, which complicates inspection. We introduce a multimodal passive remote wireless telemetry approach that utilizes different sensing modalities to address complementary inspection challenges. Polarimetric imaging successfully differentiates between typical product appearance variation and types of defects achieving an accuracy of 96% using the number of detected clusters with statistical significance at <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\alpha =0.05$ </tex-math></inline-formula> (streaks versus fibers: t(df =28) <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$= -14.062$ </tex-math></inline-formula> and fibers versus scratches: t(df =28) <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$= -7.393$ </tex-math></inline-formula>). Infrared (IR) imaging is investigated to detect glass scratches otherwise obscured by reflections, nonplanar surfaces, or vaccine powder; 100% of defects on the vial neck and body are detectable, while 3/5 smaller defects on the vial bottom can be detected. We further establish detection limits for IR imaging, showing that glass scratches with a width as small as <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$32~\mu $ </tex-math></inline-formula>m are detectable. Finally, we establish a weighted sum of pixel differences between frames to detect scratches during vial rotation. We show that this measure can be used to detect scratches at different rotation speeds, but the signal-to-noise ratio increases with rotation speed from 14:1 at <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$12~{^{\circ}}$ </tex-math></inline-formula>/s, over 8:1 at <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$24~{^{\circ}}$ </tex-math></inline-formula>/s–2.5:1 at <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$45~{^{\circ}}$ </tex-math></inline-formula>/s.

The evolution of Connected Vehicles (CVs) has introduced significant advancements in both in-vehicle and vehicle-edge platforms, creating a highly connected ecosystem. These advancements, however, have heightened exposure to cybersecurity risks. This work reviews emerging security challenges in the CV ecosystem from a new perspective, focusing on the integration of in-vehicle platforms such as the infotainment system and vehicle-edge platforms. By analyzing case studies such as Android Automotive, Message Queuing Telemetry Transport (MQTT), and the Robot Operating System (ROS), we identify the primary security threats, including malware attacks, data manipulation, and Denial of Service (DoS) attacks. The discussion extends to privacy concerns and the lack of trust-building mechanisms in CVs, highlighting how these gaps can be exploited. To mitigate these risks, we retrieve solutions drawn from the broader field of Internet of Things (IoT) security research, including Multi-Factor Authentication (MFA) and trust-based systems. The proposed framework aims to increase the trustworthiness of devices within the CV ecosystem. Finally, we identify future research directions in adaptive mechanisms and cross-domain security.

The amount of data collected for sensing tasks in scientific computing is based on the Shannon-Nyquist sampling theorem proposed in the 1940s. Sensor data generation will surpass 73 trillion GB by 2025 as we increase the high-fidelity digitization of the physical world. Skyrocketing data infrastructure costs and time to maintain and compute on all this data are increasingly common. To address this, we introduce a selective learning approach, where the amount of data collected is problem dependent. We develop novel shift-invariant and spectrally stable neural networks to solve real-time sensing problems formulated as classification or regression problems. We demonstrate that (i) less data can be collected while preserving information, and (ii) test accuracy improves with data augmentation (size of training data), rather than by collecting more than a certain fraction of raw data, unlike information theoretic approaches. While sampling at Nyquist rates, every data point does not have to be resolved at Nyquist and the network learns the amount of data to be collected. This has significant implications (orders of magnitude reduction) on the amount of data collected, computation, power, time, bandwidth, and latency required for several embedded applications ranging from low earth orbit economy to unmanned underwater vehicles.

We present a method for resolving multiple chipless RFID tags in a read zone using phase measurements conducted during a raster scan with a directive reader antenna. We differentiate between 3 chipless RFID tags with up to 2 shared resonance frequencies in the 3 - 4.4 GHz band, that are spaced as close as 6.4 cm apart. Furthermore, we automate the process of tag identification and location estimation with a positional error of less than 8 mm. We also present a relationship between tag resonant frequency and minimum tag spacing. Future directions are also discussed.

The time spent on cleaning and preparation in the operating room accounts for up to 23% of the average procedure time. Reducing these times could decrease cost and optimize care. Preparation time includes surgical instrument counting (SC) to prevent the problem of retained surgical instruments (RSI), which can cost hospitals over $200K per incident. Current SC methods are inadequate for smaller tools like surgical needles (SN). This study explores the use of a high-resolution circular chipless RFID tag for automatically differentiating suture needle sizes. The change in the signal response of the chipless RFID tag, affected by an overlaid SN, is used as a feature set to train a random forest classifier. We demonstrate that the system achieves a classification accuracy of 98.55%, effectively distinguishing between three needle sizes. These findings validate the potential of chipless RFID technology combined with machine learning as a viable tool for automated SC. Future research directions are also discussed.

This work-in-progress innovative practice paper presents a novel approach to 1) extract tacit knowledge from expert trainers while they perform a task demo, 2) decrease the learner's cognitive load via the use of instructional videos portraying the variables at play during a task demonstration, and 3) define quantifiable metrics of expertise by extracting features that differentiate experts from novice practitioners. Implicit or tacit knowledge is know-how that experts develop with experience and is difficult to verbalize, formalize or explicitly transfer to others. For this reason, knowledge transfer from expert to apprentice is usually slow and inefficient. Our approach seeks to support knowledge transfer using technology-enhanced approaches. Here, we focus on extracting and describing exper-tise. We do so by instrumenting experts, trainees and their tools with sensors that can help structure and formalize knowledge. Our first application of this framework is on the knowledge transfer between an expert and novice glassblower. Glassblowing is well known for its crucial expert/apprentice relation and its slow learning rate due in part to the difficulties in verbal transfer of skills. Our framework seeks to capture relevant data while an expert glassblower demonstrates basic actions in a beginners glass blowing course. Our sensors collect eye-tracking activity, verbal demo instructions, pipe accelerometry, air infusion, scene video and muscle activity (EMG), which continuously monitor the expert, their explanations, the tools, and the glass piece. We bring together all the sensed data into instructional videos to be used by novice learners as supportive training material. We present preliminary results related to metrics of expertise and future steps towards gathering similar data from novices. This will help develop AI-based models to extract data-driven differences between experts and apprentices, which can be used as further instructional material. We will also present plans to test the instructional effectiveness of the developed videos and how our approach can be used in other training settings involving tacit knowledge transfer.

A low-cost sensor integrating chipped ultrahigh-frequency radio frequency identification (UHF RFID) and paper flow channels is designed to characterize water content (WC) in oil. We first introduce the sensor design consisting of tag antenna, tag integrated circuit (IC), and probe with integrated paper flow channel. We then discuss the fabrication of sensor prototypes and conduct measurements of pure water, pure SAE oil, and SAE/water 70%/30% emulsion (S/W 70/30). The sensor is seeded with the fluid under test (FUT) and the sensor signal is recorded using an UHF RFID reader while the FUT propagates in the flow channel. We show that our sensor can perfectly differentiate between water, oil, and S/W 70/30 using the RSSI at discrete times as well as the slope of the RSSI-over-time signal in two distinct areas. Finally, we show that linear discriminant analysis (LDA) and affinity propagation (AP) on data transformed with principal component analysis (PCA) are effective methods for automated FUT classification in supervised and unsupervised scenarios, respectively.

Health monitoring during physical exercise is relevant to avoid issues like dehydration, specially for vulnerable population or in warm climates, since it may become critical in these cases. Current dehydration monitoring solutions are not intended for the general population, since they are expensive or require the action of wearing the device. Hence, a cost-efficient monitoring technology that could be integrated into everyday clothing would allow the democratization of eHealth, and thus, improving the health of the general population. We have been researching on low-cost sensing technologies, allowing the detection of dehydration, while maximizing the trade-off between functionality and cost. In this paper, we present a passive antenna-based UHF RFID sensor, allowing non-invasive dehydration monitoring on fabrics at low-cost. We determine that the dielectric relative permittivity (ε′ <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">r</sub> ) and loss tangent (tan(δ)) values exhibit a change up to 10 and 0.3 respectively in the presence of euhydrated (i.e. regular hydration) and dehydrated sweat in the UHF band. We use these results to design a UHF RFID tag attached to a fabric, capable of differentiating between these hydration states. After prototype implementation, we demonstrated good classification metrics at laboratory environment, reaching a 100% accuracy using K-means unsupervised learning when attempting to differentiate between euhydrated and dehydrated sweat in fabrics with concentrations over 58%. To the best of our knowledge, this combination of results is presented for the first time in the literature. Our results demonstrate the feasibility of detecting dehydration at low-cost in an unassisted manner, and thus, the possibility of democratizing eHealth for the general population.

The promise of Brain-Computer Interfaces (BCIs) is counterbalanced by concerns about vulnerabilities. Recent studies have revealed that EEG-based BCIs are susceptible to security breaches. However, current attack approaches are challenging to execute in real-world settings because they need access to, at a minimum, the EEG data stream. In this work, we introduce an unexplored vulnerability of current EEG-based BCIs that consists of remotely injecting false brain-waves into the recording device. We do this by transmitting amplitude-modulated radio-frequency (RF) signals that are received by the physical structure of the EEG equipment. We demonstrate the versatility of our system by successfully attacking three different categories of EEG devices: research-grade (Neuroelectrics), open-source (OpenBCI), and consumer-grade (Muse). We test our attack system by taking control of three different BCIs: a virtual keyboard speller, a drone-control interface, and a neuro-feedback meditation interface. Our system was successful in each case, forcing the input of any desired character with the virtual keyboard, crashing the drone, and reporting false meditative states, respectively. To the best of our knowledge, this is the first time that an EEG device is remotely hacked at the physical layer. This work shows the risks that can arise from this type of attacks, which can not only be dangerous by seizing control of a BCI, but could also lead to severe misdiagnoses in clinical EEG tests.

The EPC GEN 2 communication protocol for Ultra-high frequency Radio Frequency Identification (RFID) has offered a promising avenue for advancing the intelligence of transportation infrastructure. With the capability of linking vehicles to RFID readers to crowdsource information from RFID tags on road infrastructures, the RF-enhanced road infrastructure (REI) can potentially transform data acquisition for urban transportation. Despite its potential, the broader adoption of RFID technologies in building intelligent roads has been limited by a deficiency in understanding how the GEN 2 protocol impacts system performance under different transportation settings. This paper fills this knowledge gap by presenting the system architecture and detailing the design challenges associated with REI. Comprehensive real-world experiments are conducted to assess REI's effectiveness across various urban contexts. The results yield crucial insights into the optimal design of on-vehicle RFID readers and on-road RFID tags, considering the constraints imposed by vehicle dynamics, road geometries, and tag placements. With the optimized designs of encoding schemes for reader-tag communication and on-vehicle antennas, REI is able to fulfill the requirements of traffic sign inventory management and environmental monitoring while falling short of catering to the demand for high-speed navigation. In particular, the Miller 2 encoding scheme strikes the best balance between reading performance (e.g., throughput) and noise tolerance for the multipath effect. Additionally, we show that the on-vehicle antenna should be oriented to maximize the available time for reading on-road tags, although it may reduce the received power by the tags in the forward link.

Fueled by the growth in autonomy and the widespread availability of cellular and other forms of connectivity in transportation, interest in Vehicle-to-Everything (V2X) communication has risen remarkably in recent years. While primarily viewed as a catalyst for the future of fully autonomous driving, V2X communication holds far-reaching implications for modern transportation systems, including smart traffic lights. We present a tailored two-factor authentication scheme for verifying the identity of vehicles engaging with smart infrastructure. This scheme effectively thwarts potential malicious actors situated on roadsides from assuming the guise of legitimate vehicles, thus bolstering the security of traffic signal controllers. These controllers rely on vehicle-shared messages to ascertain real-time traffic conditions and refine signal phase timings. The proposed scheme combines a non-line-of-sight (NLOS) channel for cryptographic certificate verification with a line-of-sight (LOS) channel for visual confirmation. Diverging from prior efforts concerning the integration of NLOS and LOS authentication for Vehicle-to-Infrastructure (V2I) security, our approach capitalizes on visual attention within the infrastructure-based computer-vision system to visually validate the vehicle's identity. To demonstrate the scheme's efficacy, we implemented it on a hardware platform. Furthermore, we explore the interplay between vehicle speed, detection accuracy, and security through hardware-in-the-loop simulation.

The ability to detect and keep in lanes is crucial for the safe operation of autonomous mobile robots in construction sites and their coordination with humans in autonomous ports or logistic centers. While computer vision-based lane detection algorithms perform well under normal conditions, their performance may degrade under low visibility conditions and in adverse weather. Since robots are not constrained by human perception limits, this paper proposes a radio frequency (RF) pavement marking system that builds on Radio Frequency Identification (RFID), a short-range communication technology, to provide lane-detection assistance. We present not only the hardware designs for the RFID systems on both vehicles and roads but also a filtering algorithm to mitigate the noise in the backscattered RF signals for lane detection. Experimental results show that the information on lane keeping provided by the RF pavement markings aligns with the visual channel when mobile robots move at a speed of less than 40 miles per hour.

Rising vaccine production and complex visual characteristics of freeze-dried products have highlighted a critical need for accurate, high-speed automated quality control. Current inspection procedures, that rely on human vision or line cameras, have undesirable error rates. We propose a novel use of polarimetric imaging for defect capture and compare the performance of polarimetric imaging to RGB imaging for defect detection on vaccine vials with freeze-dried product. Vaccine vials with artificial defects (scratches and fibers) and without defects but with product appearance variations (streaks) are prepared. We capture a data set of RGB images and polarimetric images: Polarization Intensity (PI), Degree of Linear Polarization (DoLP), Angle of Polarization (AoP). We find that the differences between product variation and defects in RGB images are not statistically significant with α = 0.01 (<i>t</i>(8) = 2.088 for scratch vs. streak,<i> t</i>(8) = 2.789 for fiber vs. streak). In contrast, the differences between product variation and defects for polarimetric imaging are statistically significant for all polarization characteristics with α = 0.01 (PI: <i>t</i>(8) = 39.753 for scratch vs. streak, <i>t</i>(8) = 13.039 fiber vs. streak, DoLP: <i>t</i>(8) = 16.537 for scratch vs. streak, <i>t</i>(8) = 17.018 for fiber and streak, AoP: <i>t</i>(8) = 6.764 for scratch vs. streak, <i>t</i>(8) = 4.702 for fiber vs. streak). This indicates that polarimetric imaging may be used as a more effective technique than RGB imaging for defect detection.

A food-safe cost-effective time–temperature indicator (TTI) sensor for cold chain disruption detection at the item level is proposed. The sensor is based on the radar cross Section (RCS) readout from a chipless square split ring resonator (SSRR) exposed to organic oils with customizable melting temperatures and defined flow paths. The inclusion of several oil mixtures into the same sensor allows for the determination of a range of configurable temperatures/times. The same sensor has two modes of operation: one for threshold detection and another for gradual change detection. These modes depend on the orientation of the sensor on the packaging and the influence of gravity. The provided design, along with a convenient signal conditioning strategy, accurately detects four time exposure thresholds in the 7–30 min range when placed in upright position at ambient temperature, while it exhibits linear response between 10 and 30 min just by turning it by 90°. Prospective future directions are also discussed.

The utility of chipless radio frequency identification (RFID) technology for the identification of eight materials (seven plastic types and glass) in the recycling chain is examined. We first demonstrate how the frequency response of eight circular ring resonators (CRRs), designed to operate in the 1.5–6-GHz band, is used to encode eight unique material IDs. Tagged items are inserted one at a time in front of a chipless RFID reader and the speed and accuracy of material identification, based on the ability to decode the tag’s frequency response, is determined. We demonstrate the ability to correctly identify the material type using a random forest classifier with up to 93% accuracy while being 3– <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$16\times $ </tex-math></inline-formula> faster than other methods reported in the literature. A technique to further improve the detection speed, taking measurements with lower resolution, at the cost of accuracy is then presented. Finally, future directions of exploration are discussed.

Lane keeping is crucial for the safety of both human-driven and autonomous vehicles (AVs). Today most lane markings are either paint or reflectors. Lane recognition can be challenging in low visibility conditions when there is wear on the markings and in adverse weather such as fog or snow. Given that machines need not be limited to human perception limits, one solution is to have lane markings identify themselves to communicate with AVs through radio frequency. The RF lane markings’’ can be achieved through the use of Radio-frequency identification (RFID), a short-range communication technology, embedded in lane markings, along with RFID readers and antennas inside the car. Previous work focuses on tags in the centers of lanes to help guide vehicles. This paper explores an alternative design with tags located on the boundaries of lanes and antennas mounted on the sides of vehicles. In addition to supporting vehicle positioning, our approach can explicitly support lane departure prevention as vehicles can easily confirm if it is confined to a lane by reading tags from two edge markings. We present requirements posed by road geometry, vehicle size and motion, and the physics of radio wave propagation, and discuss how these constraints inform the design of RF lane marking systems.

Previous work on misbehavior detection and trust management for vehicle-to-everything (V2X) communication security is effective in identifying falsified and malicious V2X data. Each vehicle in a given region can be a witness to report on the misbehavior of other nearby vehicles, which will then be added to a “blacklist.” However, there may not exist enough witness vehicles that are willing to opt-in in the early stage of connected-vehicle deployment. In this article, we propose a “whitelisting” approach to V2X security, titled proof-of-travel (POT), which leverages the support of roadside infrastructure. Our goal is to transform the power of cryptography techniques embedded within vehicle-to-infrastructure (V2I) protocols into game-theoretic mechanisms to incentivize connected-vehicle data sharing and validate data trustworthiness simultaneously. The key idea is to determine the reputation and the contribution made by a vehicle based on its distance traveled and the information it shared through V2I channels. In particular, the total vehicle miles traveled for a vehicle must be testified by digital signatures signed by each infrastructure component along the path of its movement. While building a chain of proofs of spatial movement creates burdens for malicious vehicles, acquiring proofs does not result in extra costs for normal vehicles, which naturally want to move from the origin to the destination. The POT protocol is used to enhance the security of previous voting-based data validation algorithms for V2I crowdsensing applications. For the POT-enhanced voting, we prove that all vehicles choosing to cheat are not a pure Nash equilibrium using game-theoretic analysis. Simulation results suggest that the POT-enhanced voting is more robust to malicious data.