The OSCAR is six legged robot demonstrator for our organic computing project. The main objective of this project is to develop self-organizing, self-healing, and self-optimizing control architectures for robots based on organic principles. With the OSCAR robot we focus on developing a stable walking behaviour in unstructured and rough terrains, autonomous identification of failures or “injuries” of the robot, and an adaptation to environmental as well as body changes like failures or breakdowns of the robot’s components.

The firefly based pulse coupled biological oscillator concept has been successfully applied for achieving self-organized synchronization of walking robot gait patterns by dynamically prolonging and shortening of robot’s legs stance and swing phases.

The results from the experiments done on our hexapod robot demonstrator show the practical usefulness of this biologically inspired approach for run-time self-synchronizing of walking robot gait pattern parameters. Movie about this concept (45MB)...

A novel and patent pending mechanism has been introduced for robot leg amputation by joint leg walking robots.  In combination with our organically inspired approach for robot reconfiguration - SIRR (Swarm Intelligence for Robot Reconfiguration) it is used to perform in situ walking robot self-reconfiguration. Such approach can be very useful when the joint leg robots have some legs malfunctions and the legs have to be amputated. Amputating the damaged robot legs would mean that the robot will not have to carry on the malfunctioned legs during the rest of the mission. Instead, the robot will perform leg amputation, followed by self-reconfiguration and continue with its mission tasks.

The video demonstrates such self-reconfiguration done on our robot OSCAR which using the innovative mechanism amputates the malfunctioned legs, then performs the self-reconfiguration using the SIRR approach and still continues with walking.

Watch the movie about self-reconfiguring robot OSCAR (49MB)…

We have conducted innovative research on applying swarm intelligence for autonomous robot reconfiguration. Robot self-reconfiguration is significant for situations when the robot
has some failures during its operation and therefore it has to autonomously reconfigure itself in order to successfully continue the predefined mission. The motivation for this approach is to overcome the burdensome conventional design and programming of the full inverse kinematics for the robot for every possible situation. Therefore we have introduced Swarm Intelligence based approach inspired by the collective behavior of decentralized, self-organizing agents, as seen within the natural systems: bird flocking, fish schooling and bacterial growth. The SIRR (Swarm Intelligence for Robot Reconfiguration) has been successfully tested on our OSCAR robot demonstrator and it will be used in the future projects for achieving autonomous robot self-reconfiguration in cases when the robot has some failures. Watch movie about S.I.R.R. concept (36MB)...

It is very important characteristic for autonomous robotic systems to demonstrate fault tolerance. For the domain of fault tolerant robots exhibiting emergence we have been researching on developing novel methods for self-healing and self-reconfiguring robots. Therefore we have developed an organically inspired computational concept – RADE (Robot Anomaly Detection Engine) which is used to detect anomaly conditions within the robot system during its operation. RADE approach is implementing artificial immune system which is inspired by paradigms found in natural immune system. Similarly how the natural immune system works, the RADE method uses metrics to distinguish between self and non-self i.e. normal and anomaly situation. Initial tests conducted on our robot OSCAR demonstrated that it is robust method with in-situ dynamically adapting characteristic (see animation on the right). Due to its intrinsic properties for self adaptation it will be also considered to address the research towards unsupervised robot learning and autonomous robot reconfiguration.

To achieve a walking behavior in the OSCAR robot we pursue the idea of a biological theory, presenting a decentralized control architecture for controlling the walking behavior of six-legged walkers. Each leg has its own operating unit, controlling the leg lifting up and swinging to its front position and attaching back on the ground again and pushing the robot’s body to the desired direction. Communication between the several units and a very simple rule, avoiding two neighboring legs lifting up at the same time, lead to the coordination of the six legs.
The above shortly described control architecture leads to different robot walking patterns. Without a central unit and without a pre programmed walking pattern the robot starts walking with a pentapod (five feet on the ground) changing over to a tertapod (four feed on the ground) and ending in a tripod (three feet on the ground) gait. It is not defined how the gaits have to change or that they have to change. By increasing the robots velocity it adapts its gaits. This behavior emerges by the communication between the several modules. Watch movie...

Additionally we have conducted experiments on achieving equally distributed pressure on the robot’s feet during walking with means of emergence. Each leg is lifting or pushing towards the ground depending on the level of pressure on its neighboring legs. As result this leads to equalized pressure on all the feet of the robot during walking and leveling its body appropriately in respect to the ground. This emergent characteristic is useful when the robot is walking on uneven and bumpy terrain. Watch movie...