Motion Planning for Articulated Bodies with Branches

1
Motion Planning for Articulated Bodieswith
Branches

• Ilknur Kaynar-Kabul

2
Multi-branch articulated body

• Motion planning for multi-branch articulated body

Initial configuration
Final configuration
3
State of the art

• The method consists of building and searching a
  graph connecting the local minima of a potential
  function defined over the configurations pace.
• The graph is explored by means of a randomization
  technique that escapes the local minima by
  executing Brownian motions.
• This planner is considerably faster than previous
  path planners and it solves problems with many
  more degrees of freedom.

Automatic Motion Planning for Complex Articulated
Bodies, Jérôme Barraquand, 1991
4
State of the art

• In this approach, each of the sub-problem is
  solved by using numerical potential fields
  defined over bitmap-based representations of the
  2-dimensional sub-spaces.
• An efficient backtracking mechanism based on a
  novel notion of virtual forbidden regions in
  these 2-dimensional subspaces is presented.

• Practical Global Motion Planning for Many
  Degrees of Freedom A Novel Approach Within
  Sequential Framework,
• Kamal K. Gupta, Xinyu Zhu, ICRA 1994

5
Goal of the project

• Motion planning for many degrees of freedom (DOF)
  arms
• Constrained based motion planning will be applied
  to each branch
• Each branch will have its own goal configuration
• A guiding path for each branch will be found (?)
• The length of the branches will be a constraint
  in the guiding path computation
• While the branches are following their path,
  constraints will be applied for achieving the
  dynamic properties of the cable, handling the
  collisions and positional constraints (?)
• Constraints of each branch will effect the other
  branches
• The simulation may get unstable because of the
  competition of branches and constraints

6
What has been done?

• Literature survey
• Discussion of ideas with Russ
• 2 approaches
• Hierarchical motion planning
• Priority based motion planning
• Some coding

7
Hierarchical motion planning

• Apply constraint based motion planning to each
  branch
• Find paths for the leaf nodes and apply
  attraction forces to make them follow the path
• After they reach their goal configurations,
  adjust the positions of the parent nodes in the
  hierarchy from bottom-up

B
A
D
A
E
F
B
C
G
C
E
F
G
D
8
Priority based motion planning

• It is an iterative approach
• Find path for branch i and make it to follow the
  path and come its goal configuration
• Apply step 1 to branch i1, then branch i2, n
• Issues in this approach
• The assignment of priorities to the branches
• The performance of prioritized planning versus
  coordinated planning

9
Priority based motion planning

• It will be very slow for long chains
• There is no guarantee that you will be able to
  find a path for branch i1 after you adjust the
  path for branch i
• A solution to this problem can be to recompute
  the path for branch i if we cannot find a path
  for branch i1. But this makes it very slow.
• Narrow passages and complex obstacles may be a
  problem

10
Implementation

• Adaptive dynamics (AD) code for articulated body
  simulation from Stephane and Russ
• I generated the multi-branch articulated body
  using AD
• PRM Constraint based sampling Potential
  forces for goal Handling Positional Constraints
  are ready

11
What remain to be done

• It is very difficult to come up with a complete
  solution to this problem
• Current plan is to propose an approach for
  solving this problem under some assumptions
• There wont be any intersection between the paths
  of the branches
• There may be more than one solution to the
  problem
• The lengths of the branches will be sufficient to
  reach the goal configurations

12
Expected contributions by the end of the semester

• If we can find an approach that works in most of
  the cases under certain assumptions, this will
  be the first constraint based approach for
  multi-branch articulated bodies

13
Other long term goals

• The algorithm can be applied to legged robots and
  reconfigurable robots
• There might be applications such as
• Multi-branch cable route planning
• Legged robot moving on the terrain

14
Questions Ideas