The RUNES partners at Lund Institute of Technology have developed a sensor network-based mobile pendulum robot. The objectives of the project which was managed by Prof. Karl-Erik Årzén, Dan Henriksson and Dr Anton Cervin were to perform a test case for control over a sensor network and investigate the performance that can be achieved using state of the art sensor network technology such as Telos B and ZigBee.
|
|
Robot by Lund Institute of Technology.
|
Hardware Specifications
The robot comprised 2 DC motors, 1 Angle sensor, 1 Telos mote, 2 ATMEL AVR Mega8 (motors), 1 ATMEL AVR Mega16 (pendulum sensor) and a I2C bus.
Routing Scenarios
Two routing scenarios were studied. According to the first, during certain times and in certain regions the robot communicates directly with controller node 1 (single hop), otherwise the robot communicates with controller node 2 via the forward node (two hops).
According to the second scenario, the robot always communicates with both controller 1 (directly) and controller 2 (via forwarding). The control signal that is received first is taken into account, whilst control signals arriving afterwards are discarded.
Experiences
The initial experiments have shown that the transmission rate is slow (approx 10 ms per hop for 20 byte packets). If a 50 ms nominal sampling interval is used, then 1-2 hops between controller and mobile robot are feasible.
Unless the communication within the network is scheduled to avoid collisions, a very large amount of collisions leading to retransmissions or lost packets occur. Also, if the internal communication is scheduled, packet losses do occur. The consequences of closed loop control are as follows: a) the application must tolerate long time periods in an open loop, b) local backup controller co-located with sensors and actuators is required for safety-critical applications or open-loop unstable processes, and c) time stamping is essential. Telos and TinyOS worked surprisingly well however, documentation was insufficient. Finally, the implementation of fairly advanced controllers using floating point was straightforward.
Further Info
For more information contact Prof. Karl-Erik Årzén. A video that demonstrates the operation of the robot is also available.
