Project Proposal: Student: Rowan Pivetta Supervisor: Dr Nasser Asgari
The Competition 2013 Milestones Research Robotic Design Obstacle Detection Navigation Proposal
Competition run by National Instruments ◦ Supply products to engineering industry April to late September 2013 24 universities from Australia & New Zealand Flinders University team ◦ Joel Cottrell, Joshua Renfrey ◦ Rowan Pivetta and Peter Cook (BENG Mechanical) Design, build and program an autonomous robotic vehicle Fully utilises the NI Single board RIO Program the sbRIO through NI LabView ◦ A graphical development software package
Scenario: A hazardous mining environment Search and retrieval of precious metals Competition involves ◦ Integration – Robot design and sbRIO ◦ Task – Speed vs time ◦ Obstacle Avoidance ◦ Object Detection and Handling ◦ Navigation 90% of robot control must be done by single board RIO and programmed in LabView
Navigate from home position to the mining field Distinguish the golden cubes from the rubble grey cubes Pick up and transport up to 4 golden cubes to the depot Complete the task in under 3 minutes without incurring any penalties. ◦ Hitting walls ◦ Picking up rubble Mining Area Depot Home
Unknown obstacles around mining area Unknown position of the cubes in mining area Unknown openings into mining area One point for localisation ◦ A IR emitting tower with fixed co-ordinates Mining Area Depot Home ? ? ? ? ? ? ? ?
Milestone 1 – 29 th April ◦ Completion of NI LabView online training course ◦ Demonstrate control of a motor or a sensor ◦ Complete written project proposal Milestone 2 – 3 rd June ◦ Prototype design with obstacle avoidance Milestone 3 – 22 nd July ◦ Final design and demonstration of navigation and localisation Milestone 4 – 26 th August ◦ Demonstrate object handling, navigation, localisation and obstacle avoidance Competition – Week of 23 rd September
Competition run in 2012: ◦ Search and rescue mission Understand competition ◦ Grid pattern reference ◦ Height obstacles ◦ Colour block detection Check milestones ◦ The type of hardware used ◦ The sensors and structures used ◦ Assessing different designs Group discussions
Four grippers ◦ Design to maximum efficiency Internal block storage Tip-up truck ◦ Allow easier transportation Circular vs rectangular footprint Holonomic vs conventional drive Final design still under consideration Collaborative decision
Sensors are the eyes and ears ◦ Robot decisions based on sensor readings Strategic positioning ◦ Detect walls and height obstacles ◦ No interference between robotic components More sensors provides more information allowing better decision making Does to much hinder performance ◦ Increase weight and power drain ◦ Constrained to capabilities of sbRIO ◦ Further complex sensor fusing algorithms
Infra-red or Ultrasonic range finding sensors Camera can detect the edges of the obstacles LIDAR gives more information about the obstacles ◦ Can be used to determine exact boundaries Combining camera and LIDAR can distinguish between the obstacles and the smaller blocks IR Range Finder LIDAR Camera
Navigation and localisation is a difficult problem to solve in robotics Combine all sensor information Techniques ◦ Markov localisation ◦ Extended Kalman Filter Robot prediction vs measured position ◦ Simultaneous Localization And Mapping ◦ Follow the right wall
Design the robot for what is best suited for the specifications of the competition ◦ Unique wheel configuration ◦ Unique navigation solution ◦ Unique robotic design
National Instruments, 2013, NIARC, September 2012, waT1BaTSi9Xxltw&index= Bayuk, R. “A comparison of Robot Navigation Algorithms for an Unknown Goal”,