Summer Internships in autonomous Robotics (SIAR)

Developing autonomous capabilities in humanoid robots with application to robotic soccer.

This page provides an overview of the types of project available under the SIAR programme. For an introduction to SIAR and details of the application process please go here.

Indicative projects

Projects will be divided into 3 broad areas. Within each area, an indication of possible projects (to be finalised) is given below. Applicants should rank their preferences in terms of project areas.

Project area: Sensor processing and inference (SPI)

A critical part of all robotics applications is the ability to process sensor data and infer important information about the environment from the data. In the RoboCup Soccer, Standard Platform League application, the primary sensor available is the robot's camera (single colour camera with resolution up to 640x480). In addition, the robot includes inertial sensors (accelerometers and gyroscopes) and foot pressure sensors. This application motivates a number of projects in semsor processing and inference. Indicative projects (to be finalised) include:

• Implementation of Automatic Robot Location Algorithms
• Real-Time Stereo Vision Based Localisation and Object Recognition
• Obstacle detection using the GOLD paradigm
• Fast accurate robust ball tracker
• Single Camera Object Detection & Tracking for Robot Soccer
• Line feature detection algorithm development from webcam images

Project Area: Balance Control and Action (BCA)

Another core autonomous robotics skill involves actuation. In the case of wheeled robots, this is relatively straightforward where geared motors may be controlled to give desired motions. In the case of legged robots, particularly humanoid types, there is much work required to plan the correct sequence of joint movements to create a 'good' walk, to be able to kick the ball, and to be able to balance when disturbed (for example by accidental collisions with other objects). Example projects in this area include:

• Balancing a planar robot using arm motion.
• Robot tilt stabilization based on foot pressure feedback
• Robot kick design
• Goal keeper saves

Project Area: Learning, Planning and Co-ordination (LPC)

To make the best use of the sensor information and the available actuators, intelligent, strategic decision making is required. This allows the robots to cooperate as a team, to make best advantage of their possible actions, to give the team the best chance of both avoiding opponent goals, and scoring their own goals. Possible projects in this area include:

• Capture & Analysis of Opposition Data to Improve Game Performance
• Irregular Cellular Automata for Robot Motion Planning
• Multi-criteria path planning & decision-making for co-operating robots
• Specification and Simulation of Soccer Tactics.