The Ray Research Institute presented a new video of its motorcycle robot that is being trained to perform jumps, flips and maneuvers reminiscent of professional parkour athletes, the project is called UMV and the idea behind it is different from anything we have seen so far, instead of choosing between robots with legs or robos with wheels, the researchers decided to combine both, the efficiency of a motorcycle with the jumping ability of a bipedal robot.

The result is a hybrid machine capable of traveling on flat terrain with speed, but also of jumping over obstacles, climbing structures and adapting to extreme environments. The most impressive thing is not only the hardware, but the way it learns. This robot was not programmed manually for each movement, it was trained with reinforcement learning, a type of artificial intelligence where the machine learns by trial and error in millions of simulations, each action is rewarded or penalized and over time the robot discovers for itself how to move in the most efficient way possible.

The result is almost surreal, the UMV manages to jump obstacles of more than 1 m, perform somersaults in the air, maintain balance on a single wheel and adapt its movements in real time, as if it were thinking while moving, and there is a detail that makes all this even more important, this type of mobility is not just for demonstration, it is the basis for robots in environments where humans have difficulties, uneven terrain, natural disasters, risk areas and extreme exploration.

When we put this together with everything we have seen, robos that think, that work, that overcome physical limits, it becomes clear that evolution does not occur in isolated parts, it is converging intelligence, body, movement, adaptation, all at the same time and perhaps the most impressive thing is this, we are no longer teaching machines only to execute tasks, we are teaching them to move in our world, not as we do, but better than we do and when machines start to live in the real world, they need to understand exactly how they move.