Webinar – Train a Balance Bot with Reinforcement Learning

Reinforcement learning (RL) is one of the most exciting areas of modern artificial intelligence (AI), enabling robots to learn complex behaviors through interaction and experience rather than traditional programming. Combined with physics simulation, RL allows robots to master skills such as balancing, walking, and recovering from falls in a virtual environment pre-deployment to real hardware.

As powerful as these technologies are, getting started can be challenging. Understanding the tools, workflows, and deployment process often presents a steep learning curve for developers and robotics enthusiasts alike.

In this hands-on workshop, Shawn Hymel, Expert Instructor and Creative Course Creator in Embedded Systems and Machine Learning, will guide participants through the complete process of training a self-balancing robot using reinforcement learning and deploying the trained model to real hardware.

Using the M5Stack Bala-C or Bala2 Fire platforms, attendees will learn how to:

• Import a 3D model of the M5Stack Bala-C into the MuJoCo physics simulator
• Train a neural network policy using a multi-phase PPO (Proximal Policy Optimization) curriculum
• Deploy the trained actor network to an ESP32 microcontroller using Arduino
• Understand the sim-to-real pipeline and how learned behaviors transfer from simulation to physical robots

Beyond the mechanics of training and deployment, the workshop explores a foundational workflow increasingly used in modern robotics development. By leveraging simulation before hardware testing, developers can accelerate learning, reduce risk, and create more capable autonomous systems.

Whether you're a robotics hobbyist, embedded developer, student, or AI practitioner, this session offers a practical introduction to the tools and techniques powering the next generation of intelligent machines. You'll leave with a deeper understanding of reinforcement learning, robotics simulation, and the complete path from training an AI model to running it on a real robot.

When: Thursday, August 13, 2026, 10:00 AM CDT

Length: 1 hour

Register here to attend or receive the full recording at any time following the event: https://event.on24.com/wcc/r/5384134/958EFFE494BF1C6A2AD94F9B093DC01B?partnerref=blog