Smart behavior for real scenarios

Navigating a vehicle requires not just planning the detailed route to go from start to destination, but also requires anticipating the many things that could potentially happen along the immediate trajectory over the next several seconds of travel, and controlling the vehicle moment to moment in such a way that it can respond safely to those various possibilities. Good human drivers do it. Autonomous vehicles will have to do it even better.

Our motion planning technology gives AV stack partners the ability to “imagine” all the potential things that could happen and their various probabilities, as identified by their sensors and AI-powered perception stack, and calculate the best immediate motion plan that ensures safety in balanced anticipation of those possibilities. Unlike most systems that can only consider a single forecast at a time, our dedicated technology, known as Lightning, can run through hundreds of potential forecasts per sensor cycle. The ability to consider far more potential outcomes in less time means the vehicle can maneuver much more safely and efficiently. 

Lightning enables AV stack partners the ability to program for specific vehicle dynamic characteristics and capabilities under various conditions, as well as for custom decision matrices.



Using a customers perception system we can compute a collision-free plan in under a millisecond. Using information about the scene (obstacles and their various probable future trajectories) we can examine and determine the best path based on various partner-determined cost factors.

1 path / agent



We exploit your inputs about agent probable future paths and perform a computation that adjusts automatically to agent motions and environment changes as they occur. Such plans can, e.g., slow down and create some space when driving past an unattended child, in case the child runs into the road, or start overtaking a car but retreat if it subsequently speeds up.

1+ path / agent


Reactive + Risk-Aware

This is an advanced form of computing and executing autonomous vehicle behavior in real-time based on  think-ahead predictions of where other agents in the environment are traveling.

1+ path / agent

Want more?

Fill out this form and we will get you more details on how we can enhance your autonomous solution