The Digital Twin Campus Lab (“DTCL”), established by SoftBank Corp. (“SoftBank”) and Keio Research Institute at SFC (“KRIS”), announced it launched a proof of concept (PoC) for the advanced operation of autonomous shuttle buses utilizing digital twin technology at Keio University Shonan Fujisawa Campus (“SFC”), located in Fujisawa City, Kanagawa Prefecture, in May 2023.
In this PoC, DTCL collects information on moving objects from LiDAR sensors and traffic signal lights from cameras installed on the rooftops of buildings for the Digital Twin Campus Lab Platform (“DTCL Platform”). KRIS is utilizing the collected information to research the advanced operation of the autonomous shuttle buses, which is being conducted in collaboration with Kanagawa Chuo Kotsu Co., Ltd.
One challenge facing the operation of autonomous shuttle buses, is a limited range of detection due to relying solely on sensors and cameras installed onboard. By replicating information from sensors installed in buildings on the DTCL Platform, for example, and sharing that information with autonomous shuttle buses, SoftBank and KRIS aim to enhance autonomous shuttle bus operations and realize comfortable and safe transportation.
This PoC is one part of the research and development initiative conducted by the DTCL starting from October 2022. The research focuses on collaboration between physical and virtual spaces to identify issues, solve problems, and develop technologies such as position estimation.
(1) Detection of oncoming vehicles during right turns
There are cases where it is difficult to recognize vehicles approaching from a distance by relying only on the sensors of autonomous shuttle buses. In such situations, by sharing information from sensors installed on the rooftops of buildings through communication, it is possible to complement the areas that autonomous shuttle buses cannot recognize on their own and expand the areas recognized by the buses.
In this PoC, SoftBank and KRIS enabled the real-time acquisition of information on oncoming vehicles at right turn points within SFC from the DTCL Platform. Previously, drivers manually performed right turns after visually confirming the absence of oncoming vehicles. However, by obtaining information from the DTCL Platform, it is possible to advance to operations where buses automatically turn right if there are no oncoming vehicles, thus enhancing the level of automation.
(2) Comfortable and safe vehicle operation through signal light prediction
Signal light information is crucial in the operation of autonomous shuttle buses. For example, by predicting when a signal light is about to turn red, the bus can decelerate in advance, improving passenger comfort and safety. However, detecting signal light information using cameras installed on buses can present challenges, such as difficulties in detection due to backlighting.
In this PoC, SoftBank and KRIS performed signal light estimation using AI based on imagery from fixed cameras capturing signal lights in the vicinity of SFC. This estimated information can be obtained from the DTCL Platform. Additionally, based on historical signal light data, it is possible to predict how soon signals will change. By integrating this information with autonomous shuttle buses, it becomes feasible to achieve operational service that is safer and more comfortable.