Towards Generating Simulated Walking Motion Using Position Based Deep Reinforcement Learning.
William JonesSiddhant GangapurwalaIoannis HavoutisKazuya YoshidaPublished in: TAROS (2) (2019)
Keyphrases
- reinforcement learning
- humanoid robot
- real robot
- initial position
- human walking
- human gait
- position information
- position and orientation
- function approximation
- relative position
- motion analysis
- image sequences
- motion model
- human arm
- optical flow
- coordinate frame
- motion planning
- motion estimation
- human motion
- space time
- walking speed
- reference frame
- multi agent
- state space
- precise localization
- angular velocity
- markov decision processes
- motion segmentation
- motion field
- camera motion
- machine learning
- learning algorithm
- computer vision
- upper body
- gait analysis
- dynamic programming
- reinforcement learning algorithms
- motion tracking
- model free
- motion capture
- motion patterns
- optimal policy
- motion parameters