Humans or machines are faster? RC cars racing with reinforcement learning
HUN-REN SZTAKI researchers already raced their RC car in Japan, but at the event on 9 November 2024, the forces between machine and human were also measured in Budapest.
Two dozen participants tested their reflexes against the power of algorithms in a research competition organised at the Tódor Kármán College. The contestants were able to follow the camera mounted on the car with a so-called First Person View (FPV) headset, and control the RC car with a steering wheel and pedals. All the while, the autonomous car used the Reinforcement Learning (RL) algorithm, already proven in Japan: car learns from the consequences of its own decisions.
From the results of the race, the researchers concluded that the machine is quite skilful, easily outperforming average drivers, but there is still chance to beat it. Most of all, by someone with more experience and better driving skills than the average driver.