Biped Robot P1 that can tolerate any maltreatment from humans

Chinese startup LimX Dynamics has proven the ability of robots to quickly respond to any emergency situations and confidently navigate difficult terrain. The design is based on a self-learning model and the ability to respond to external hazards, including moving objects or uneven ground.

LimX Dynamics uses a separate model to train artificial intelligence algorithms. In a new test of the platform, the company placed the Biped Robot P1 on Mount Tanglang in Shanghai. The company tested the system’s ability to navigate in an unknown environment with rough terrain.

The testing required Biped Robot P1 to remain constantly aware of objects around the robot and then feed it to a reinforcement learning system. The robot confidently climbed grassy hills, navigated ditches, and recovered its gait after stumbling. P1 was even able to remain stable after repeated blows to his limbs with a stick.

Reinforcement learning uses rewards or penalties to allow AI to make autonomous decisions. Such algorithms differ from classical artificial intelligence systems in that they are based on autonomous learning. The startup now aims to apply similar technology to its CL-1 humanoid robot, which can process data from stairs and other objects before successfully navigating them.