Improving the safety of bicycle riders is one of the critical issues for Autonomous Driving. The crossing intention of the cyclist is expected to be predicted from the onboard camera of autonomous vehicle. In a real traffic situation, a cyclist usually turns his or her head to check the situation of the back of him or her before he or she crosses the road. Therefore, the action of turning head is an important signal to indicate the intention of crossing a road. This research proposes to detect the behavior of the turning head based on the body and head orientation using deep neural networks. This research is awarded by Consumer Electronics Society Japan Chapter.

Autonomous vehicles keep evolving in years. Pedestrian detection is a significant function for autonomous vehicles. However, there is an issue that affects the performance of pedestrian detection, which is the lighting condition. Pedestrian detection performance usually decreases at nighttime. This research utilized Deep Neural Network and multispectral images to improve the performance of pedestrian detection in various lighting conditions.

The Augmented Reality (AR) navigation system can provide a new experience for pedestrians compared to the conventional navigation on 2D map. This research presents an AR based indoor navigation system. The proposed system adopts Simultaneous Localization and Mapping (SLAM) to build a point cloud map, and performs positioning and navigation in a novel hybrid map which integrates the 3D point cloud map and floor map of the indoor environment. 

Stock performance prediction is one of the most challenging issues in time series data analysis. Machine learning models have been widely used to predict financial time series during the past decades. Even though automatic trading systems that use Artificial Intelligence (AI) have become a commonplace topic, there are few examples that successfully leverage the proven method invented by human stock traders to build automatic trading systems. This study proposes to build an automatic trading system by integrating AI and the proven method invented by human stock traders. 

*Research was initially conducted at Kamijo lab of the University of Tokyo

Image based human behavior and activity understanding has been a hot topic in the field of computer vision and multimedia. As an important part, skeleton estimation, which is also called pose estimation, has attracted lots of interests. For pose estimation, most of the deep learning approaches mainly focus on the joint feature. This research proposes a novel multi-task framework for the human pose estimation. The proposed framework integrates the joint feature, body boundary, body orientation and occlusion condition together for human pose estimation.

*Research was initially conducted at Kamijo lab of the University of Tokyo

Drive Data Recorder (DDR) stores the relevant driving data to provide feedback on driver behavior for accident analysis and insurance issues. Conventional DDRs are standalone devices, record many useless data or lose important information. This research proposes to develop intelligent DDR in smartphones to replace conventional DDR products. 

*Research was conducted at Kamijo lab of the University of Tokyo

Accurate vehicle self-localization is significant for autonomous driving. This research proposes a low-cost passive sensor based vehicular positioning system which can achieve lane-level performance in urban canyons. This research proposes to employ 3D building map information to analyze the satellite signals and improve the accuracy of GNSS positioning. In addition, this research further proposes to integrate GNSS receiver with other passive sensors: Inertial Measurement Unit (IMU), vehicle speedometer and onboard camera. The proposed system can achieve 95% correct lane recognition rate and sub-meter accuracy in various tests in Tokyo city. 

*Research was conducted at Kamijo lab of the University of Tokyo