MIT Robotics - Gaurav Sukhatme - Learning Coordinated Performant Flight with 20 Neurons

MIT Robotics · 61:40

Gaurav Sukhatme (USC) shows that end-to-end reinforcement learning, trained only in simulation, can teach tiny Crazyflie quadrotor swarms to fly. The drones learn to set raw rotor speeds, reach goals, avoid each other...

Read the full summary on tuber

Redirecting...