Join us in an exciting challenge and build an autonomous robot to push the limits of what low-cost, open-source hardware and deep learning on the edge can do for humanity. We are excited by a world in which drones, rovers, robots and underwater ROVs help us in our daily lives - delivering food and medicine to those in need, improving agricultural techniques, responding to emergencies and much more.
We want to see what autonomous bots you can dream up using advanced hardware kits, sensors, computer vision, and deep learning based on Arm technologies.
Your project should make use of autonomous vehicle or drone technology based on Arm, follow the theme of this contest (autonomous bots) and be able to achieve at least one of the following tasks:
- Autonomously transport a package in an urban, rural or underwater environment
- Autonomously assist in a real-world scenario
Your innovative machine can be a drone, rover, underwater ROV or vehicle of your own making. Here are a few of Arm’s favorite technologies to incorporate in your design:
- A Raspberry Pi controlled DonkeyCar. Optionally apply for a free DonkeyCar kit ($200 value) by submitting your idea before July 21st or go here to find a complete bill of materials to purchase one on your own
- Pixhawk 4 flight controller
Early applicants will be eligible to receive one of the 50 4-wheeled DonkeyCars that we're distributing to developers who submit compelling proposals around how these small wheeled robots can be used for good.
We are giving away thousands of dollars to lucky winners in the form of robots and other cool objects! Our judges at Arm are going to pick the best qualifying projects based on the judging criteria outlined in the rules section.
Post a fun and entertaining video on social media with the hashtag #ArmRobotChallenge and you could win a GoPro HERO 6!
There are a few additional ways to earn bonus points from our judges.
1. Earn up to 10 bonus points for a project that showcases a real use case in your local environment, providing improvements to your community. Does your bot deliver food or aid to those in need? Report aerial data back from a remote location? Show us your bot in action!
2. Address one or more of the following themes in your project and earn up to 10 bonus points per theme:
AI on the Edge (10 pts)
- Does your machine use the latest frameworks released by Arm to enable deep learning on Arm processors?
- Does it make the best use of the MCUs on board? Is it capable to use the Cortex-M processors for inference?
Sustainability (10 pts)
- Does your solution help people to live more healthy, sustainable and peaceful lives?
- Solar Power/Charging does your robot take advantage of renewable energies. Does it park in the sun to recharge?
- Does your solution help to conserve energy?
Mapping (10 pts)
- Able to create a map of an unknown environment and use it to navigate
- Mapping disaster zones to help first aid / deliver supplies - maybe your DonkeyCar delivers emergency supplies, or your homemade robot maps lava flows to identify open escape routes
There are major communities in support of all the devices required for this contest. Example projects on getting started with a Raspberry Pi can be found here. Search Hackster’s project hub to find tutorials on more complicated topics like computer vision and deep machine learning. The DIY Robocars community has also created some great projects that you can access from here.
Watch our on-demand technical webinar where Will Roscoe and Adam Conway, co-founders at Donkey Car, walk you through how to use the Raspberry Pi with the Donkey Car open source Python library to build a completely autonomous rover.
Want to get started with artificial intelligence on the edge?
- Use the software development kit released by Arm. Arm NN bridges the gap between existing NN frameworks and the underlying IP. It enables efficient translation of existing neural network frameworks, such as TensorFlow and Caffe, allowing them to run efficiently – without modification – across Arm Cortex CPUs and Arm Mali GPUs.
- Use Compute Library to develop AI applications on Cortex-A processors and GPUs, taking advantage of highly optimized libraries that will boost the performance of your robot.
- Use the CMSIS-NN kernels or uTensor framework to implement inference on your Arm Cortex-M processors.
Want to submit as a team? Go for it! You’re welcome to divide and conquer with a team of up to 5 members. But remember, one prize per team.
The Arm team and other participants are accessible for help and questions as well via the contest discussion forum.
Good luck and happy building!