snake-like reinforcement learning assignment.txt

snake-like-reinforcement-learning:
Your player controlled character is in a 1000x1000 unit board.
The character has a squint angle of 90 degrees, meaning if facing north, it can
see 45 degrees on either side of north. It can estimate distance of the reward.
The character is always moving forward at a speed of S = 100units/sec
The character turns with a turn rate of 30 deg/sec (max) and can turn slower if desired.

Somewhere randomly on the board is a "goal".
Train a model using reinforcement learning such that the character repeatedly
arrives to the goal using a relatively optimal path.

Arriving: within 5 units (radius) of the reward.

visual representation as a gif