Automated human motion in constrained environments

1
Automated human motion in constrained environments

• Maciej Kalisiak
• mac_at_dgp.toronto.edu

2
Introduction

• human character animation
• constrained environments
• kinematic method
• currently 2D, extendible
• sample solution

3
Path Planning

• piano movers problem
• given start and goal configurations
• find connecting path

4
Application to Human Motion
5
Approach

• starting point RPP
• additions
• moving while in contact with environment
• notion of comfort
• knowledge of human gaits

6
Understanding RPP

• Randomized Path Planning
• a path planning algorithm

7
Simplest Planner

• characters state q
• repeated perturbations,i.e., Brownian motion
• repeat until goal reached

8
Building a Potential Field

• discretize into grid
• potential Manhattan distance to goal
• flood-fill

9
Gradient Descent

• character ? point mass
• sample qs neighbourhood
• pick sample with largest drop in potential
• iterate until goal reached
• not feasible analytically

10
Local Minima

• gradient descent stops at any minimum
• use random walks to escape
• Brownian motion of predetermined duration
• use backtracking if minimum too deep
• revert to a previous point in solution,followed
  by a random walk

11
Deep Minimum Example
12
Smoothing

• solution embodies complete history of search
  process
• also very noisy
• a trajectory filter post-process is applied
• removes extraneous motion segments
• makes remaining motion more fluid

13
Modifications

• grasps and grasp invariants
• comfort heuristic system
• gait finite state machine
• grasp-aware gradient descent, random walk,
  smoothing filters

14
Character Structure

• 10 links
• 9 joints
• 12 DOFs
• frequent re-rooting

15
Grasp Points

• represent potential points of contact
• three types
• reduce the grasp search space
• summarize surface characteristics

16
Grasp Invariants

• each gait dictates
• the number of grasps
• the types of grasps
• enforced by the GFSM
• rest of planner must not alter existing grasps

17
Motion without Heuristics
18
Heuristic System

• each heuristic measures some quality of q
• D(q) overall discomfort, a potential field
• getting comfy gradient descent through D(q)

19
Implemented Heuristics
20
The Gait FSM

• states represent gaits
• each edge has
• geometric preconditions
• motion recipe
• priority
• self-loops gait-preserving motion that changes
  grasps

21
Complete System
22
More Results
23
Future Work

• 3D
• quadrupeds, other characters
• grasp surfaces
• non-limb grasping
• add concept of time, speed
• use machine learning

24
FIN
http//www.dgp.toronto.edu/mac/thesis
25
Appendix

• (extra slides)

26
Alternate gradient descent view
27
Smoothing Algorithm
28
Need for Limb Smoothing
29
Limb Smoothing Solution
30
Implemented GFSM
31
Contributions

• human character animation algorithmfor
  constrained environments
• grasp point discretization of environment
• grasp constraint
• comfort modeling using heuristics
• gait FSM
• adapted RPP algorithms to grasp constraint