A robot will be used for reconnaissance in an urban environment and must use GPS and other sensors to determine its position and the path it will follow
Department: School of Systems Engineering, Systems Engineering
Supervised by: William Harwin
Mobile robots must localise their position with respect to their environment. This is achieved by integrating information from a range of sensors such as an on board GPS (global positioning system), a magnetic compass, odometers, and two laser range finders. The difficulty is that erroneous information from any one sensor may severely bias the results. Often this is because there is a poor or incorrect characterisation of the sensor data. This summer placement will investigate two aspects of mobile robot navigation, the consistency of the sensor data and consequent movements of the robot. An opportunity exists within this project to contribute results from this project directly to the University's entry into the MOD challenge event that will run in the second half of August (www.ruchallenge.com). The first goal is to investigate the sources of noise or inconsistency in the sensor data stream. In the case of the GPS sensor the key element is the number and location of the satellites that are used to make the estimate. Data from the laser range finders will already be processed and the goal will be to see if this information can be used to further improve the consistency of the robot location. Testing the sensor reliability will need to be done in context of the robot path and the second goal of this summer placement will be to investigate smooth robot paths based on splines or polynomials. A series of test courses can then be arranged and the consistency of the robots movement across the course can be evaluated against the sensor data.
The student must log the information from GPS, camera and compass data over extended periods of time. A typical log would be upto an hour of data with the robot static, followed by the robot following a rectangular course returning to its initial position. This task should last about 2 weeks. A further two week period will be needed to do a detailed analysis of the log data and to identify the primary causes of inconsistency and to revise the approach to path planning. A two week run up period is envisaged culminating in the University entry into the MOD challenge, and during this period the student will work closely with the existing team and Thales UK who are coordinating the T3 entry. The final two week period will be used to tidy up the approach and draft a short technical document. If results are sufficiently encouraging this would also form the basis for a conference based publication.
The student should have a good background in electronics with strong analytical skills. Some computing experience will also be needed. The ability to use the Matlab package is a prerequisite. Experience with micro electronics communication protocols such as the CAN bus would be advantageous although not necessary.
The student will develop greater analytic skills. He or she will also gain exposure to current techniques in robot state estimation such as the Kalman filter, and use this knowledge to investigate the consistency of information that can be extracted from multiple sources of data. Since some researchers have postulated that humans may use similar techniques to model our interaction with the world the student will also be exposed to the literature in these areas. The student will also gain experience in team working, particularly during the run-up stage to the MOD challenge.
UoR
Monday 30 June 2008 - Friday 29 August 2008
Applications should be via email with a c.v. attached. The applications should be made to w.s.harwin@reading.ac.uk.