Transcript lis-5-99

Visual Robot Navigation



Low-level behaviors are built in or tuned with
reinforcement learning
Avoid obstacles using optical flow
Fixate and drive to a distant object
Map starts as a graph of views with behaviors on
the arcs
With further experience
Aggregate different views of the same place
Incorporate metric information
4/12/2016
1
Experimental Set-Up





Robot works in same virtual environment as
human subjects
No need for the head-mounted display!
Easy to simulate robot motion
Most development work in software-only
environment
Validation runs on real robot wearing head tracker
4/12/2016
2
Current Progress



Obstacle avoidance, wall-following using optical
flow from textured objects in virtual scene
Simple histogram-based view representation and
matching
Learn a route in the environment
Human drives robot through corridors with
joystick
Robot finds sequence of its behaviors that are
most consistent
Stores route as views connected by behaviors
Can re-create the route
4/12/2016
3
Technical Issues
Representation and matching of views
 Pixel arrays vs. histograms
 2D vs. 3D
Aggregation of views into places
 Role of odometry
 Role of 3D geometry
Role of metric information
 Angles and distances on arcs vs. real 2D
embedding
4/12/2016
4