Concept: Researchers from Baidu Research Robotics and Auto-Driving Lab (RAL) and the University of Maryland have developed a driverless Autonomous Excavator System (AES). The system can perform material loading tasks without any human intervention, replicating the performance close to that of an experienced human operator.
Nature of Disruption: AES employs real-time perception, planning, and control algorithms, as well as a new architecture for autonomous operation. The perception module integrates many sensors, including LiDAR, cameras, and proprioceptive sensors, as well as complex algorithms, such as a dedusting neural network, to perceive the 3D environment and identify target materials. The AES architecture can be used by excavators of all sizes due to its modular nature. To test AES, researchers collaborated with a leading equipment manufacturer to deploy the system at a waste disposal facility, a hazardous real-world setting where automation is in high demand. Despite the difficult task, AES was able to operate for more than 24 hours without the need for human intervention. It has also been tested in cold weather conditions. The volume of materials excavated by a compact excavator, both wet and dry, was 67.1 cubic meters per hour, which is comparable to the performance of a traditional human operator.
Outlook: Excavators are vital for infrastructure construction, mining, and rescue applications. Baidu aims to offer a secure and robust platform integrated with AI and cloud capabilities to the construction industry. To fulfill this vision, Baidu is partnering with a few construction machinery companies to automate traditional heavy construction machinery with AES. During the trial phase, AES performed continuously and reliably as compared to a human operator. Moreover, the system is useful in a harsh environment where human capabilities are limited. RAL has plans to refine core modules of AES and further explore scenarios where extreme weather or environmental conditions may be present.