A multi-layer approach of interactive
path planning for assisted manipulation in Virtual Reality
Simon Cailhol, Philippe Fillatreau, Jean-Yves Fourquet and Yingshen Zhao Ecole Nationale d’Ing´enieurs de Tarbes-Laboratoire G´enie de Production ´
Context
Industrial motivation
While designing products with the Product Lifecycle management (PLM) approach, the tasks
involving manipulations by a human operator can be simulated in Virtual Reality (VR) applications.
One of the main issues for most of such tasks is to find path for systems components to move.
Path finding tools
Off-line simulation
Path planners from robotic research field.
Computation time for highly integrated systems.
&
On-line simulation
Manipulation in VR assembly applications.
Environment hard to handle for human operator.
.
Assisted VR assembly
Use an automatically planned trajectory to guide human operator immersed in VR application.
Proposition
Human operator Environment model
Geometrical map Topological map
Semantic map
Environment modeling Path planning
Interaction means
B1,5 B2,7 B1,6 B2,8
B3,5 B3,7 B4,6 B4,8
B3,9 B3,10 B4,9 B4,10
P1 P2
P3 P4
P5 P7
P8 P6
P9 P10
P3P3 P3 P3 P4P4
P4 P4
cha racterization
cha racterization geometrical placement
cell decomp osition costs co mputation query definiti on
strategy defi n it ion
strategy defi n ition guidance
manipulations intent p rediction
re-planning triggering
semantics util ization
Contribution
Multi-layer environment model
not complex not complex
complex complex
very complex very complex
very complex complex
cluttered not complex
fixed fixed
fixed
moving B
1,5B
2,7B
1,6B
2,8B
3,5B
3,7B
4,6B
4,8B
3,9B
3,10B
4,9B
4,10P
1P
2P
3P
4P
5P
7P
8P
6P
9P
10P
3P
3P
3P
3P
4P
4P
4P
4• Semantics characterizes places and obstacles.
• Topology represents places, borders and their connectivity.
• Geometry describes obstacles and free space.
Planning process
S G
B
1,5B
2,7B
1,6B
2,8B
3,5B
3,7B
4,6B
4,8B
3,9B
3,10B
4,9B
4,10P
1P
1P
1P
2P
3P
4P
5P
7P
8P
6P
9P
10P
3P
3P
3P
3P
4P
4P
4P
4P
2P
2S G
M M M M
M M
M M M
M
A two phases path planning process involving both semantic information :
1. Coarse planning defines a path in the topology and split it in steps. Semantics controls the topology exploration.
2. Fine planning defines the geometrical path crossing each step. Semantics chose the planning technique.
Interaction and control sharing
G C
C
−
→ g
M M M M
M M
M M M
M C
C C
C C C C
C C C
C C C C
C C C G G
−
→ g
M M M M
M M
M M M
M C
C C C C
M M
M
M
M M C
M M C
• Operator’s actions are used to define Semantics utilization for cost functions and planning techniques
• Operator intent is predicted at topological level and used for path re-planning
• Authority is shared between automatic planner and human operator for geometrical manipulation
Results
Path planning 3D environment
round
pentagon triangle
square
cluttered cluttered not complex
complex very
complex