2020 Fiscal Year Research-status Report
Driver behavior modeling and its application to a guidance-as-needed steering system for individualized lane change assistance
Project/Area Number |
19K20318
|
Research Institution | The University of Tokyo |
Principal Investigator |
王 正 東京大学, 生産技術研究所, 特任助教 (20837497)
|
Project Period (FY) |
2019-04-01 – 2022-03-31
|
Keywords | Haptic interface / Driver behavior modeling / Lane change assistance / Shared control / Guidance-as-needed |
Outline of Annual Research Achievements |
Our research focuses on modeling lane change behavior of individual driver and designing a guidance-as-needed steering system through haptic interface. In FY 2020, our research achievements are summarized as follows: 1. A personalized driver decision model for discretionary lane-change initiation based on reinforcement learning was developed through a driving simulator experiment with a number of subjects. 2. A surface Electromyography (sEMG) controlled steering system was proposed for pedestrian collision avoidance through a rapid lane change task, and its performance was evaluated by comparing with a haptic guidance steering system. 3. An adaptive haptic guidance steering system via forearm sEMG measurement was designed, and a driving simulator experiment with a number of subjects was conducted to evaluate its effect on driving performance. The results indicate that the adaptive authority of haptic guidance (haptic guidance decreases when driver grip strength increases) yielded lower driver workload and reduced the lane departure risk compared to manual driving and fixed authority haptic guidance.
|
Current Status of Research Progress |
Current Status of Research Progress
3: Progress in research has been slightly delayed.
Reason
1. Based on the driver model developed in the last fiscal year, a new driving simulator experiment with a number of participants was conducted to collect the discretionary lane-change decision data, which are used to train a novel reinforcement learning model, and thereby a personalized lane change model was developed. 2. By real-time measuring driver eye movement or forearm muscle activity, adaptive haptic guidance systems were designed to assist individual drivers for lane change tasks. The results from driving simulator experiments with a number of participants demonstrate that the designed systems are adaptive and capable to improve driving safety and comfort. 3. The real-vehicle experiment to evaluate the proposed adaptive haptic guidance system was designed, while it was not smoothly carried out as planned due to COVID-19.
|
Strategy for Future Research Activity |
In the next fiscal year, our research activities will focus on the following aspects: 1. Real-vehicle experiments will be conducted to evaluate the effectiveness of adaptive haptic guidance steering system on lane change performance by comparing with fixed authority haptic guidance and manual driving. Moreover, the effect of adaptive haptic guidance on distracted driver behavior will be investigated. 2. The driver behavior measured in real-vehicle experiments will be analyzed for better understanding of driver's integration of visual and haptic information during a lane change task.
|
Causes of Carryover |
In this fiscal year, we have focused on system design and evaluation by driving simulator experiments. The real-vehicle experiments were postponed due to COVID-19. In the next fiscal year, we will conduct real-vehicle experiments by recruiting subjects. In addition, we will publish several journal papers to report our results, and therefore, budget for publication fee will be needed.
|
Research Products
(6 results)