Westwell Collaborates with Tongji University to Present Innovation on Autonomous Driving Perception at IROS 2024

Westwell Collaborates with Tongji University to Present Innovation on Autonomous Driving Perception at IROS 2024

Blog Article

From October 14 to 18 this year, one of the top-tier academic conferences on robotics and intelligent systems worldwide, the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), is held at ADNEC Centre, Abu Dhabi. In collaboration with the world-renowned Tongji University, Westwell released a research paper on the innovation of autonomous driving perception, which has been selected and officially published by IROS with widespread attention from attendees.

The Westweller shared insights on this research paper at IROS

As one of the largest and most impacting robotics research conferences worldwide, IROS gathers top researchers, scholars, and industry leaders from around the world every year. This year's theme is "Robotics for Sustainable Development", highlighting the role of robotics in achieving sustainability goals.

At the meeting, Westwell and Tongji University released a research paper in the field of autonomous driving perception systems titled “3D Object Detection via Stereo Pyramid Transformers with Rich Semantic Feature Fusion”.

This paper was jointly researched by Westwell’s autonomous driving R&D team and Professor Chen Guang from Tongji University, proposing a new method for 3D object detection based on binocular stereo images. This method is to extract features from binocular images with a novel Stereo Pyramid Transformer (SPT), which can enhance the accuracy of object detection through rich semantic feature fusion.

Westwell and Tongji University jointly released the latest research paper, which was presented at IROS 2024

Extensive experiments on the KITTI and self-collected datasets have demonstrated that this method surpasses all existing Transformer-based binocular 3D detection methods and has realized the leading CNN-based 3D detection performance for the first time. The method will also enhance the reliability of active safety systems in autonomous driving. Especially, under various weather and lighting conditions, the method can timely detect obstacles and other vehicles in the scene, and accurately perceive spatial positions and distances, thus improving driving safety. Furthermore, it can accurately sense the traffic situation on the road and optimize vehicle routes to improve operational efficiency.

This collaboration marks another important achievement in the industry-university-research cooperation between Westwell and Tongji University in the cutting-edge technology fields of "AI, Autonomous Driving, and Embodied AI", receiving high recognition from top global academic experts and industry professionals. This partnership showcases the leadership of Westwell and Tongji University in technological innovation and their commitment to driving global sustainable development.

In the future, Westwell will continue to focus on New AI and New Energy, facilitating a smarter, greener evolution of large-scale logistics that is more efficient and sustainable, while actively promoting global productivity.