Presented by David Stavens. Autonomous Inspection Compute a path such that every point on the boundary of the workspace can be inspected from some point.

Slides:



Advertisements
Similar presentations
Great Theoretical Ideas in Computer Science
Advertisements

A* Search. 2 Tree search algorithms Basic idea: Exploration of state space by generating successors of already-explored states (a.k.a.~expanding states).
Traveling Salesperson Problem
Weighted graphs Example Consider the following graph, where nodes represent cities, and edges show if there is a direct flight between each pair of cities.
Approximations of points and polygonal chains
 Distance Problems: › Post Office Problem › Nearest Neighbors and Closest Pair › Largest Empty and Smallest Enclosing Circle  Sub graphs of Delaunay.
Lecture 24 Coping with NPC and Unsolvable problems. When a problem is unsolvable, that's generally very bad news: it means there is no general algorithm.
Approximation Algorithms for TSP
1 The TSP : Approximation and Hardness of Approximation All exact science is dominated by the idea of approximation. -- Bertrand Russell ( )
NUS CS5247 Motion Planning for Camera Movements in Virtual Environments By Dennis Nieuwenhuisen and Mark H. Overmars In Proc. IEEE Int. Conf. on Robotics.
Data Structure and Algorithms (BCS 1223) GRAPH. Introduction of Graph A graph G consists of two things: 1.A set V of elements called nodes(or points or.
© The McGraw-Hill Companies, Inc., Chapter 8 The Theory of NP-Completeness.
Generated Waypoint Efficiency: The efficiency considered here is defined as follows: As can be seen from the graph, for the obstruction radius values (200,
Ant Colony Optimization. Brief introduction to ACO Ant colony optimization = ACO. Ants are capable of remarkably efficient discovery of short paths during.
Contour Stitching Introduction to Algorithms Contour Stitching CSE 680 Prof. Roger Crawfis.
Chapter 3 The Greedy Method 3.
1 Last lecture  Configuration Space Free-Space and C-Space Obstacles Minkowski Sums.
Complexity 15-1 Complexity Andrei Bulatov Hierarchy Theorem.
1 Discrete Structures & Algorithms Graphs and Trees: II EECE 320.
Visibility-based Motion Planning Lecture slides for COMP presented by Georgi Tsankov.
Approximation Algorithms: Combinatorial Approaches Lecture 13: March 2.
Randomized Planning for Short Inspection Paths Tim Danner and Lydia E. Kavraki 2000 Presented by David Camarillo CS326a: Motion Planning, Spring
Motion Planning with Visibility Constraints Jean-Claude Latombe Computer Science Department Stanford University.
3 -1 Chapter 3 The Greedy Method 3 -2 The greedy method Suppose that a problem can be solved by a sequence of decisions. The greedy method has that each.
MAE 552 – Heuristic Optimization Lecture 27 April 3, 2002
1 Target Finding. 2 Example robot’s visibility region hiding region 1 cleared region robot.
CS 326A: Motion Planning Criticality-Based Motion Planning: Target Finding.
CSE 326: Data Structures NP Completeness Ben Lerner Summer 2007.
MAE 552 – Heuristic Optimization Lecture 26 April 1, 2002 Topic:Branch and Bound.
Randomized Planning for Short Inspection Paths Tim Danner Lydia E. Kavraki Department of Computer Science Rice University.
Rendezvous Planning in Mobility- assisted Wireless Sensor Networks Guoliang Xing; Tian Wang; Zhihui Xie; Weijia Jia Department of Computer Science City.
Online Data Gathering for Maximizing Network Lifetime in Sensor Networks IEEE transactions on Mobile Computing Weifa Liang, YuZhen Liu.
9-1 Chapter 9 Approximation Algorithms. 9-2 Approximation algorithm Up to now, the best algorithm for solving an NP-complete problem requires exponential.
CSE 550 Computer Network Design Dr. Mohammed H. Sqalli COE, KFUPM Spring 2007 (Term 062)
Randomized Planning for Short Inspection Paths Tim Danner and Lydia E. Kavraki 2000 Presented by Dongkyu, Choi On the day of 28 th May 2003 CS326a: Motion.
Probabilistic Roadmaps for Path Planning in High-Dimensional Configuration Spaces Kavraki, Svestka, Latombe, Overmars 1996 Presented by Chris Allocco.
Probabilistic Roadmaps for Path Planning in High-Dimensional Configuration Spaces Lydia E. Kavraki Petr Švetka Jean-Claude Latombe Mark H. Overmars Presented.
Hidden Surface Removal
Visibility Graphs and Cell Decomposition By David Johnson.
Chapter 5.4 Artificial Intelligence: Pathfinding.
Shortest Path Algorithms. Kruskal’s Algorithm We construct a set of edges A satisfying the following invariant:  A is a subset of some MST We start with.
Programming & Data Structures
1 Introduction to Approximation Algorithms. 2 NP-completeness Do your best then.
© The McGraw-Hill Companies, Inc., Chapter 3 The Greedy Method.
The Traveling Salesperson Problem Algorithms and Networks.
© Manfred Huber Autonomous Robots Robot Path Planning.
1 The TSP : NP-Completeness Approximation and Hardness of Approximation All exact science is dominated by the idea of approximation. -- Bertrand Russell.
Mehdi Kargar Aijun An York University, Toronto, Canada Discovering Top-k Teams of Experts with/without a Leader in Social Networks.
Graph Theory Topics to be covered:
1 Introduction to Approximation Algorithms. 2 NP-completeness Do your best then.
Energy-Aware Scheduling with Quality of Surveillance Guarantee in Wireless Sensor Networks Jaehoon Jeong, Sarah Sharafkandi and David H.C. Du Dept. of.
Great Theoretical Ideas in Computer Science.
Ch. 6 - Approximation via Reweighting Presentation by Eran Kravitz.
Probabilistic Roadmaps for Path Planning in High-Dimensional Configuration Spaces (1996) L. Kavraki, P. Švestka, J.-C. Latombe, M. Overmars.
Approximation algorithms for TSP with neighborhoods in the plane R 郭秉鈞 R 林傳健.
CSE 589 Part VI. Reading Skiena, Sections 5.5 and 6.8 CLR, chapter 37.
Search CPSC 386 Artificial Intelligence Ellen Walker Hiram College.
Weighted Graphs Computing 2 COMP s1 Sedgewick Part 5: Chapter
1 An Arc-Path Model for OSPF Weight Setting Problem Dr.Jeffery Kennington Anusha Madhavan.
1 Energy-Efficient Mobile Robot Exploration Yongguo Mei, Yung-Hsiang Lu Purdue University ICRA06.
CSC 413/513: Intro to Algorithms
Navigation Strategies for Exploring Indoor Environments Hector H Gonzalez-Banos and Jean-Claude Latombe The International Journal of Robotics Research.
Motion Planning Howie CHoset. Assign HW Algorithms –Start-Goal Methods –Map-Based Approaches –Cellular Decompositions.
1 Euler and Hamilton paths Jorge A. Cobb The University of Texas at Dallas.
BackTracking CS255.
Example robot cleared region robot’s visibility region hiding region 2
CS 326A: Motion Planning Probabilistic Roadmaps for Path Planning in High-Dimensional Configuration Spaces (1996) L. Kavraki, P. Švestka, J.-C. Latombe,
Last lecture Configuration Space Free-Space and C-Space Obstacles
Haitao Wang Utah State University WADS 2017, St. John’s, Canada
Clustering.
Presentation transcript:

Presented by David Stavens

Autonomous Inspection Compute a path such that every point on the boundary of the workspace can be inspected from some point on the path. Can think of this as a watchman’s route. Key Innovation: Handles sensors with limited range and visibility (incidence).

Incidence Constraint θ must be less than θ max.

Background on Watchman Problem Chin and Ntafos, 1988: – O(n 4 ) in simple polygons. – But NP-Hard if the polygons have holes. Hoffmann et. al. (1998) gives online, constant- factor approximation algorithm.

One Out-of-Date Assumption? They assume sensing “is a time consuming operation” thus “it makes sense to identify at which points on the inspection path it is actually necessary to perform the sensing.” This assumption seems out-of-date with modern robots. Sensing for this application is computationally cheap (eg: SICK Lidar).

Randomized Algorithm Expands Gonzalez-Banos and Latombe [1998]. Pick point P from the set of points on the border of the workspace that aren’t guarded. – (Efficient implementation might use trees.) Region R that can see P (equals what P sees) is constructed and clipped according to sensing limitations.

Randomized Algorithm (cont’d) Region R is sampled K times. Each sample is evaluated as a new guard. Sample that guards most NEW length is kept. – (Would backtracking be an improvement?) Repeat above until everything is guarded.

Algorithm Illustration Selected Point Already Covered

Algorithm Illustration Selected Point Already Covered Visibility Region (Note we must check orientation in this case – not discussed in algorithm. The sensor is NOT omnidirectional.)

Algorithm Illustration Selected Point Already Covered Visibility region Two Sensing Candidates

Algorithm Illustration Newly Covered Area Already Covered Visibility region Keep Candidate Seeing More New Boundary

Solving TSP Connecting the guards is the well-known Traveling Sales(wo)man Problem (TSP). – TSP is NP-Complete. Fortunately: Simple 2-approximation exists. – Use the pre-order MST walk. – Works provided graph obeys triangle inequality: A -> B + B -> C ≥ A -> C

Building the Graph Authors use the shortest-paths graph. Fully connected graph: – One edge for each pair of guards. – O(N 2 ) in total. Edge weight for (i,j) is length of the shortest path between (i,j). How do we compute this?

Computing Shortest Paths Start with a SECOND graph containing: – Nodes: All guard positions and workspace vertices. – Edges: Any free straight line between two nodes. – Weights: Euclidean distance Then search the graph. Optimization: Only connect nearby points. – Minimal effect since MST uses short edges.

2D Result 1

2D Result 2

2D Result 3

What About 3D? Algorithm works “largely unchanged.” But authors want an optimization to avoid “visibility volumes,” which are expensive to compute. – Recall these are used to pick and evaluate guard locations. Key: Optimize separately for “visible surface enumeration” and to determine where to sample.

Determining Where to Sample Range constraint is a sphere. Incidence constraint is an (infinite) cone. Thus: – Calculate sphere-cone intersection. – Ray-trace between workspace boundary point and points on the intersection to verify visibility.

Visible Surface Enumeration …allows us to determine what surfaces a guard location candidate can see. Uses very basic algorithm from graphics: – Iterate over surfaces, front-to-back order. – Obviously, surfaces occlude those behind them. Many ways to deal with this, paper doesn’t give great depth on this subject.

One Problem Iterating can have circular dependences!

3D Result 1

3D Result 2

Closing Remark Partially Observable Markov Decision Processes (POMDPs) are another way to generate policies that inspect or clear spaces. Can even make guarantees about intruder not being able to “slip by” the robot. – Doesn’t seem to be the case here.

POMDP Example – Will show file-not-available if streamed. – Must be saved. Work by Joelle Pineau, McGill University Formerly part of Thrun’s lab at CMU.