MindVLA can also navigate complex traffic environments by jointly modeling its own vehicle behavior and the trajectories of other vehicles. It employs RL to learn from mistakes, continuously optimizing its decision-making.

→ View original post on X — @jiqizhixin,