Groundwork Toward the Inaugural AI Formula

Issa Omura･Masaya Okada･Yuki Akimoto･Kota Sakazaki･Atsushi Kato･Yuji Yasui (Honda R&D)

In light of the increasing industry demand for autonomous driving and AI technology in recent years, SICE and JSAE are introducing the AI Formula technical challenge wherein student teams will compete in their design of a fully autonomous driving system. The AI Formula challenge provides an excellent opportunity for rising engineers to acquire the skills and technology necessary for next-generation mobility research. This paper details our groundwork toward the inaugural AI Formula.

Vision-Based Control for Autonomous Driving Systems in AI-Formula

Shogo Hoshino･Akito Nemoto･Phearamony Phan (Gunma University)･Shunsuke Tarui･Akichika Kurihara･Yuki Asaga･Takahiro Oinuma･Yuji Komoto･Ryuki Kondo･Tatsuki Note (Kogakuin University)

In the AI-Formula project of Honda Research and Development, we are conducting research and development of an automatic driving system. In this presentation, we will explain the details of the information acquisition process using cameras and the control methods applied to it. Future research issues and prospects will also be discussed.

Optimization of price plan for a smart satellite city with price sensitivity considered

Shota Zenke･Zhenlong Wu･Takehito Kobayashi･Wenjing Cao (Sophia University)

Due to the declining birthrate and aging problem, there are concerns about future shortage of transportation options. In this study, aiming for continuous operating of mobility services using EVs, we introduced price sensitivity into a smart satellite city simulator and calculated the base fee and distance fee that maximize annual revenue using dynamic programming. Simulation results showed that taking price sensitivity into account leads to increased revenue.

Exploration of Solutions for the Benchmark Problem of Autonomous Emergency Avoidance through Multiple Approaches

SIYANG XIE･TAKAYASU KUMANO･YUJI YASUI (Honda R&D)

Autonomous emergency avoidance systems play a crucial role in establishing a safe driving environment. In this study, we developed a benchmark problem based on predefined autonomous emergency avoidance scenarios. Multiple research teams devised solutions using their unique approaches, which were compared and evaluated in a competition format. The evaluation results are presented, and the effectiveness of each approach is discussed.

Polynomial regression-based vehicle trajectory tracking control with vibration suppression

Kazuki Ogawa･Takeru Goto･Kosuke Toda (Honda R&D)

This study proposes a control system of vehicle trajectory tracking by calculating the steering angle using polynomial regression. To improve tracking stability, the evaluation function of the regression model incorporates suppression terms for steering angle velocity and acceleration, as well as an L2 regularization term. We compared our method with general model predictive control and demonstrated its effectiveness in reducing computation time and deriving practical tracking accuracy through computer simulations. Furthermore, we validated the method with real vehicle experiments.