Towards reducing human efforts in autonomous driving system

Release Time:2024-05-22Number of visits:10

Speaker:  KaiCheng Yu, Westlake University.

Time:       4:00 pm, May 23rd

Location: SIST1A 200

Host:        Xuming He


Although GPT-based Al models have achieved widespread popularity, their virtual nature restricts their ability to reduce human effort in physical tasks, highlighting the importance of robots in realizing genuine AI democratization. In the past, we have deployed over 700 autonomous vehicles for delivery, and we have it identified that there are two major challenges preventing further scaling up. Firstly, expert intervention is necessary maintain perception modules that contain neural networks. To this end, we made several progresses in minimizing human effort through neural architecture search. Additionally, we recognize the need for 3D data annotation proportional to deployment scale. Our solution is to develop an autolabing system with a simple yet robust framework that fundamentally tackles the failure of multi-sensors setup. In closing, we present our latest effort, AutoML Perception V1, which is a broad-based autoML AI system designed to address both of the aforementioned challenges in an industrial setting. Deployed on a cloud platform, this system reduces human effort on average by 95% while significantly outperforming baseline models and drastically lowering costs.


Dr. KaiCheng Yu, Assistant Professor at Westlake University, is the principal investigator of Autonomous Intelligence Laboratory (AutoLab). He completed his undergraduate studies at the University of Hong Kong and proceeded directly to a Ph.D. program at École Polytechnique Fédérale de Lausanne (EPFL) in Switzerland. His research interests encompass automated machine learning, computer vision, 3D perception, autonomous driving, among others, with over a dozen publications in prestigious international journals and conferences. In 2019, he was awarded the Qualcomm Innovation Fellowship (one of only four recipients in Europe that year), and in 2021, he joined DAMO Academy's Autonomous Driving Lab through the Alibaba Star talent program. In 2023 and 2024, respectively, he was selected for Hangzhou's and Zhejiang Province's High-Level Overseas Programs. Prior to returning to China, Dr. Yu held R&D positions at Intel Intelligent Systems Lab and Abacus.AI, where he led or was a core member in various projects. Additionally, as a co-founder, he established KMind, a company that secured tens of millions in investment and was honored in 2023 with the title of Key Recommendation for Leading Talent Project in Yuhang District.