Changes

Added

Added LocalController class which allows users to directly connect controller to JetBot

Adds a small fix to enable later revision of waveshare JetBot.

This release is the first to support JetBot 2G as well as facilitate docker.

JetBot SD card image based on JetPack 4.3 (required for B01 revision of Jetson Nano Developer Kit ):

https://drive.google.com/open?id=1G5nw0o3Q6E08xZM99ZfzQAe7-qAXxzHN

Object Detection Engine:

https://drive.google.com/open?id=1KjlDMRD8uhgQmQK-nC2CZGHFTbq4qQQH

SD Card image

Found here: jetbot_image_v0p3p0.zip

• PyTorch 1.0 pre-installed
• TensorFlow 1.13 pre-installed
• 4GB swap enabled
• default power model 5W
• Jupyter Lab service installed
• Status display service installed

Python package

• Heartbeat class to determine disconnects from browser
• Camera class to generate eventful camera frames
• Robot class for controlling robot locomotion
• Motor class for event based motor control
• ObjectDetector class for detecting objects using the subpackage jetbot.ssd_tensorrt
• TRTModel class for abstracting TensorRT model execution on numpy arrays

Notebooks

Robot

• basic motion teaches programmatic robot control
• teleoperation demonstrates video streaming and gamepad control
• collision avoidance
  • data collection to collect free and blocked labels
  • live demo to demonstrate simple wander / avoid obstacles
• object following
  • live demo to demonstrate multi network COCO object following and simultaneous collision avoidance

Host

• collision avoidance
  • training to demonstrate training image classification network with PyTorch