AI TRAINING LAB · CANVAS2D · TOP-DOWNRBT-08 · 2D

RBT-08

Robot Parkour (2D)

Robotics · Top-down Reinforcement-Learning Training View

ROBOTICS

2D TOP-DOWN ARENAEPISODE 001

POLICY PPOSEED 1701

STEP 0CANVAS2D ACTIVE

EPISODE PROGRESS0 / 600

LEARNING TELEMETRY

Reward0.00reward

Success0%goal rate

Loss1.000policy loss

Entropy100%exploration

REWARD CURVE

CONTROL

Speed1.0× Robots24

TASK · DIFFICULTY 6/10

Train a fleet of parkour robots to weave through scattered obstacles toward the two glowing goal beacons, minimizing collisions while the policy converges. Drag on the arena to nudge the view; scroll to zoom.

+1.0 goal-0.3 collision

← Robot Parkour hub 3D version →