A new system called IMPACT, built by USC Viterbi School of Engineering and USC Stevens School of Computing and AI graduate students Karan Owalekar and Yiyang Ling, teaches robots to decide which collisions are safe rather than avoiding all contact. The system was recently accepted into the 2026 IEEE International Conference on Robotics and Automation (ICRA). The work was co-supervised by assistant professor Daniel Seita and assistant professor Erdem Biyik. IMPACT uses context to reason, without training, that a plastic bottle, harmless alone, becomes unsafe to touch when sitting next to a glass one. Owalekar noted that "the glass bottle is risky, but then a plastic bottle next to a glass bottle is equally risky, because if you hit a plastic bottle, it will hit the glass bottle, so the relative positions of everything matter." The system breaks from "collision-free" motion planning, the standard approach that trains robots to avoid all contact. Seita said the approach "breaks away with standard motion planning and robotics assumptions."