If you create your own project, please email us at deepraceropensource@amazon.com and we’d be happy to feature it on the Projects list.
By Chris Miller, CEO of Cloud Brigade
Use object detection models to identify objects and fire at them using a motorized toy blaster mounted on DeepRacer. It extends the DeepRacer hardware with a 3D-printed, servo-controlled rotating turret and custom bracket. An Arduino microcontroller is used to control the rotation servo, blaster flywheels, and ammo feeder motor over a USB/Serial link. A second camera can be added to the turret to allow DeepRacer to scan 360 degrees for objects and fire at them.
Project ideas: Sending a robot to put out a fire, multi-tasking (navigate and fire a nerf cannon), competitive racing, strategic games.
Device needed: AWS DeepRacer or AWS DeepRacer Evo
Add-on: Toy Blaster, Servo Controlled Turret, Arduino Microcontroller, 2nd Camera (optional)
By Jochem Lugtenburg, Developer at Relive, 2019/2020 DeepRacer Championship Cup finalist
Mimics a real-world car that starts and stops at traffic lights and stop signs. The logic for identifying different colors in traffic signals and detect stop signs was developed by combining various computer vision capabilities, including OpenCV image-processing functions and object-detection machine learning models. Currently the focus is on stop signs and the traffic lights. Possibilities for extension include training the car to stop when a person or animal walks in front of it and detecting and staying within the bounds of a lane. Possibilities for scenarios with multiple DeepRacers on a track include driving slower if another DeepRacer is in front of the car or overtaking other cars using the LIDAR sensor.
Project ideas: Multi-racer competition, autonomous vehicle, using AI for safety and perception
Device needed: AWS DeepRacer or AWS DeepRacer Evo
Add-on: N/A
By Martin Paradesi, Organizer of AWS Machine Learning Community: NYC meetup
Community project that leverages OpenCV image-processing capabilities to detect a mouse in the image from an infrared camera mounted on the DeepRacer. The infrared camera allows the RoboCat to detect the mouse in the dark and scare it away using a pre-configured action space with 3 actions (move forward if a mouse is detected, move back after previous action, and remain stationary).
Project ideas: Pest control, object tracking, and in-home safety
Device needed: AWS DeepRacer or AWS DeepRacer Evo
Add-on: ELP Camera