We can make drones and robots do whatever we tell them to. We’ve seen Jibo order a pizza, the humanoid robot Pepper sell iPhones, and Flirtey use drones to deliver medical supplies.

But, essentially, those are all pre-programmed functions, right? To make robots smarter and more autonomous, we need to get away from the concept of using smartphones as the brains of robots. As Lux analyst Maryanna Saenko told me, “Let’s get away from the smartphone on wheels. That’s a really limiting approach.”

And NVIDIA is trying to do just that with its Jetson TX1 module, a credit card-sized supercomputer that uses machine learning in hopes of building smarter autonomous robots. Devices that use the TX1, NVIDIA says, can learn from the environment to navigate safely, identify and classify objects of interest, and perform 3D mapping and scene modeling.

Jesse Clayton, NVIDIA’s product manager of autonomous machines, joined the Robotics Trends Show to discuss why robots need to be smarter, the challenges involved in doing that, and how the Jetson TX1 can help.