This paper presents the implementation and evaluation of a driver steering model that reflects individual driving characteristics by integrating model predictive control (MPC) with a personal modeling error corrector (PMEC). A baseline driver model is first generated using MPC applied to an equivalent two-wheel vehicle model. The deviation between the MPC-based trajectory and the actual driver's trajectory is regarded as an individual characteristic and is modeled using the PMEC. Based on driving data collected from a hardware-in-the-loop (HIL) simulator equipped with a physical steering wheel, the proposed model is identified and its accuracy is assessed through numerical simulation. The model is then deployed in the HIL simulator to reproduce driver steering behavior. Experimental results demonstrate that the proposed model more precisely captures personal driving preferences, such as lane-change timing, compared to MPC alone. Furthermore, the reproduced trajectories closely align with ideal simulations, confirming the HIL simulator as an effective platform for evaluating personalized driving support systems.

Detecting lameness in dairy cattle is essential to mitigating the effects of a significant animal welfare and health issue for the dairy industry across diverse farming systems. The locomotion score, a standard method for lameness evaluation, depends on visual assessments conducted by experienced and skilled observers, which limits the objectivity of diagnoses on large-scale farms. In this paper, we propose a gait anomaly detection method through motion reconstruction, which is based on identifying low-dimensional subspaces derived from normal motion patterns within the training data. Specifically, we first generate smooth motions from the trajectories of each key point extracted by a skeleton extraction network in a video. We then use a self-expressive model to learn subspaces from normal motions and reconstruct the given test motion. Our reconstruction module leverages the insight that as the subspace-based approximation strategy only enables reproducing the normal motions, the anomalous motions would induce a significant reconstruction error. Experimental results using cattle gait dataset demonstrate the effectiveness of the proposed method through quantitative and qualitative evaluation.

A strain generated in the chassis connected to a rover wheel can be used to estimate the rover's locomotion condition on rough terrain, such as the lunar or planetary surface. The transmission of motor power to the ground is impeded by loose soil covered with regolith, making it difficult to directly evaluate the performance of each wheel. To address this issue, strain gauges are used to measure deformation in the chassis, allowing quantification of changes that occur when there is a discrepancy in the traveling conditions between the front and rear wheels. The chassis strain data differs between the case where all wheels operate under the same conditions and the case where each wheel encounters different traveling states. Therefore, the rover's overall traveling condition can be estimated using a set of strain data collected from the wheel chassis.

This study investigates the safety evaluation of service robots in public spaces, addressing both traditional physical risks and broader concerns such as algorithmic bias and unfair user treatment. Conventional safety assessment methods primarily focus on mitigating physical risks. However, service robots, which interact with diverse users, may inadvertently exhibit biased behaviors that are difficult to detect using traditional approaches. To analyze these issues, we employ a virtual case study of a library guide robot, assessing collision risks and fairness concerns using STAMP/STPA, a risk assessment method suitable for interactive systems. The results indicate that this approach successfully identifies risk factors contributing to biased behavior.

Small and medium-sized food factories often have problems — not only in the area of “manufacturing,” such as limited automation, heavy reliance on manual labor, and production losses, but also in “integrated manufacturing and sales,” including order and delivery management, materials, personnel and equipment management, as well as manufacturing planning and execution, and so forth. In addition, they have significant issues in “product development,” such as responding to customer needs and enhancing brand strength. Given these circumstances, there is a pressing need for public support organizations to address the problems across these three core factory activities — “manufacturing”, “integrated manufacturing and sales”, and “product development” — from a holistic perspective. However, comprehensive support addressing all three domains in an integrated manner appears limited. To solve these problems, the Technical Research Institute of the Japan Society for the Promotion of Machine Industry has developed the Cross-Industry Collaboration Team Method. This approach provides team-based consulting by specialists in the fields of “manufacturing”, “integrated manufacturing and sales”, and “product development”. The team includes experts from universities, academic societies, professional organizations, public research institutes, engineering consultancies, food machinery manufacturers, and manufacturing sectors not yet engaged in food machinery. The method supports the continuous improvement of small and medium-sized food factories while creating new business opportunities for food machinery makers and related industries. This paper outlines the challenges facing small and medium-sized food factories, explains the Cross-Industry Collaboration Team Method, and reports on verification experiments assessing its effectiveness.

This paper applies round-addition differential fault analysis to the lightweight cipher CHAM, recovering full secret keys with few additional rounds: 2 for CHAM-64/128, 1 for CHAM-128/128, and 2 for CHAM-128/256.

We applied round addition DFA to LEA, and with two fault injections, successfully derived the secret key for LEA-128 and LEA-192, and partially derived 192 bits of the secret key for LEA-256.