Autonomy Agent - Stick Balance Training Mode
Activation (Actor)
Learns how to apply the correct force to the base of the pendulum (pushing left or right).
Evaluation (Critic)
Learns to evaluate the progress of the trial and predict how close the system is to failure.
HTML5 and JS version recreated by Dr Michael Woodley & Gemini (c) 2026.
Original 1986 Reinforcement Learning model by Charles Anderson.
Original NeXTSTEP Balancer.app by David J. Besemer.
Trial Number
1
Time (s)
0.0
Best Time
0.0
Engine Controls
Network State