Projects / Ideas / Jetson robot that learns to avoid collisions

Jetson robot that learns to avoid collisions

AdvancedEdge AIIdeaReviewed
In short

How do you build a jetson robot that learns to avoid collisions?

A Jetson Orin Nano on a robot base collects 'free' and 'blocked' camera photos, trains a small classifier and drives around without bumping into things. You need NVIDIA Jetson Orin Nano Super Developer Kit, IMX219 camera for Jetson Nano and Orin, Robot car chassis kit 4WD, TB6612FNG Motor Driver, Li-Po 3S 11.1 V 2200 mAh 30C (drone), XL4015 5 A buck converter module and Arduino Nano.

Level
Advanced
Time
About 600 min
Parts
7
An idea, not a guided build yet. Goal, parts and steps; no wiring, code or tests. Copy it to My Garage to plan it.
Goal

Collect a dataset, train a model with transfer learning and run it live.

How to build it

  1. Take 200 photos each of free paths and blocked paths
  2. Train ResNet-18 for two classes in PyTorch on the Jetson
  3. Export to TensorRT
  4. Drive forward when free, turn when blocked
  5. Send motor commands to the Nano over USB serial

Related projects

Common questions

What parts do I need for the jetson robot that learns to avoid collisions?

NVIDIA Jetson Orin Nano Super Developer Kit, IMX219 camera for Jetson Nano and Orin, Robot car chassis kit 4WD, TB6612FNG Motor Driver, Li-Po 3S 11.1 V 2200 mAh 30C (drone), XL4015 5 A buck converter module and Arduino Nano.

How long does the jetson robot that learns to avoid collisions take to build?

About 600 minutes for a advanced maker.