ITSS Nagoya Chapter

We introduce in general the different activities at Takeda Laboratory. Our group covers rich and diverse research fields, including autonomous driving, driving behavior, and others, all based on data-centered processing approach.

Research topics of our group cover ego-vehicle localization, traffic scene recognition, especially pedestrian attribute recognition, driving behavior prediction, and pedestrian-vehicle interaction modeling.

We use statistical signal processing and machine learning for analyzing and modeling driving behavior. Our research topics include detection of risky surrounding vehicles and driver stress estimation.

Our environmental recognition technologies using various sensors are presented, such as parking scene reconstruction by a millimeter-wave radar and self-localization by visible and thermal cameras.

Visual explanation is one of the important topic in computer vision. For self driving car, it is also important function. We proposed Attention Branch Network for visual explanation and applied it to self driving car. We explains the method and show the results in simulation environment.

We are working on a social project to realize an effective, efficient, and sustainable Data-Oriented Regional Traffic Safety Management. We extract potential blacksopts and evaluate treatments based on advanced vehicle probe data.

This video introduces our smart mobility researches on human behavior analysis, modeling, and its application to the human behavior predictive controller to realize human-friendly motilities.

Our laboratory is researching the real-time monitoring of a driver's state. We have realized the estimation of the driver's state by analyzing biological signals while driving.

The hands perform an important role in daily activities, including driving. We propose a driver monitor system based on hand activities for recognizing driver's behavior and detecting hazards. Hand activity can be detected as acceleration or vibration signals by using wrist-worn accelerometers. From the measured signals, using the vibration features and machine learning techniques, we can detect the driver's behaviors, including distracted driving activities, such as using a cell phone, navigation systems, etc. We have also shown that the driver's mental workload can estimate using the time-frequency analysis of vibration signals of the hand. Our approach using wearable sensors, such as wrist-worn accelerometers, have the potential to detect driver behavior more accurately and at a much lower cost. Currently, we are working on estimating the driver awareness and understanding the context before/after driving using the vibration features of the hand and the vehicle, then applying it to safe driving support.

For the purpose of deployment of an automatic driving system as a solution of social difficulties, we pursue research on functional features and operation of an automatic driving systems, through field trial tests of various mobility services supposed to be an social solution. Even now, the services using automatic driving are very feasible in a park, commercial facility and designated sightseeing area where are separated from public road systems and are applied specific traffic rules and regulations. We have been performing field tests such areas assuming as an early adaptation of automatic driving system and services.

This study proposed using a driver agent in the form of a compact communication robot to encourage safer driving by helping drivers recognize their own behaviors.

We research intelligent robotics technologies. This video shows our recent results on autonomous mobile robots, person identification and attending, and human-robot interaction and collaboration.

Saya is a CGI character famous for its realistic appearance. Our team develops Multimodal-Agent "Saya" integrating various sensors and AI technologies. Please check out what she can do today.