E190Q

1
E190Q Project IntroductionAutonomous Robot
Navigation

• Team Member 1 Name
• Team Member 2 Name

2
Preliminary Project Presentation

• Problem Definition
• Written definition
• Overview image
• Provide performance metrics
• Background
• Include 3 references
• Be sure to provide full citation
• Use images from references
• Describe key findings of paper

3
Preliminary Project Presentation

• Proposed Solution
• Block Diagram including sensors and actuators
  (inputs, outputs, closed loop )
• Measurable Outcomes
• List potential plots or tables of performance
  metrics
• Milestones
• List major tasks with dates
• Identify team member responsible if applicable

4
Preliminary Project Presentation

• Notes
• 5 minute time limit for slides
• Both students must present
• Students will help with assessment
• Presentations on Monday, April 1, 2013

5
Problem Definition

• To design a Multi AUV Task Planner that considers
  kinematic constraints

6
Problem Definition

• To design a Multi AUV Task Planner that considers
  kinematic constraints

7
Problem Definition

• Given
• N task point locations and M AUVs
• Determine
• The assignment of tasks to AUVs and AUV tours of
  assigned task points that minimizes the maximum
  path length all AUV tours.

8
Problem Definition

• Performance Metrics
• Maximum AUV tour length
• Planning Time or run time complexity

9
Background

• 1 R. Zlot, A. Stentz, M. B. Dias, and S.
  Thayer, Multi-robot exploration controlled by a
  market economy, in Proc. IEEE Conf. Robotics and
  Automation, vol.3, Washington, DC, pp. 3016-3023,
  2002.
• Used an auction based method in which task points
  are auctioned off to robot with the highest bid
  (i.e. lowest additional path cost).
• Decentralized.
• Fast, O(MN), but Sub-optimal

10
Background

• 2 L. E. Dubins, On curves of minimum length
  with a constraint on average curvature and with
  prescribed initial and terminal position and
  tangents, American J. Mathematics, vol. 79, no.
  3, pp. 497-516, Jul. 1957.
• Demonstrated the shortest path between points
  when minimum turn radius is a constraint
• Shortest Path is a connected curve of minimum
  radius, straight line segment, and curve of
  minimum radius

11
Background

• 3 Chow, Clark, Huissoon, Assigning Closely
  Spaced Targest to Multiple Autonomous Underwater
  Vehicles, Journal of Ocean Engineering, Vol. 41-2
  2007.
• Algorithm considers vehicle dynamics and currents
• Demonstrated that using euclidean distance
  between task points is a poor metric for
  calculating tour path length when task points are
  tightly spaced
• Real Ocean Deployments

12
Background

• 3 Chow, Clark, Huissoon, Assigning Closely
  Spaced Targest to Multiple Autonomous Underwater
  Vehicles, Journal of Ocean Engineering, Vol. 41-2
  2007.
• Algorithm considers vehicle dynamics and currents
• Demonstrated that using euclidean distance
  between task points is a poor metric for
  calculating tour path length when task points are
  tightly spaced
• Real Ocean Deployments

13
Proposed Solution
N Task Point Locations
Task Assignment Algorithm
Task Sequence Algorithm
AUV Path Construction Algorithm
M AUV Paths
M Task Assignments
M Task Sequences
M AUV Locations
14
Proposed Solution

• Task Assignment Algorithm
• Cluster N points into M groups K-means clustering
  algorithm
• Assign one AUV to each cluster using a greedy
  assignment algorithm
• Task Sequence Algorithm
• Find next closest point algorithm
• AUV Path Construction Algorithm
• Fit arc path segments between each task point of
  a sequence

15
Measurable Outcomes

• Run time as a function of the number of robots
• Average AUV path length for various ratios of N/M
• Comparison of average AUV path length when using
  standard MTSP planner and MTSP planner that
  considers kinematic constraints

16
Milestones
Data Task
Jan 15 Develop multi-AUV simulator
Feb 1 Implement Auction Based Task Planner MTSP solution
Mar 1 Implement Auction Based Task Planner MTSP solution
Mar 8 Run 100 simulations for each parameter setting
Mar 15 Present planner and results