Research plan 2008-2010

Task 1: Multi-wheel system control

Hydraulic and electric actuators with computer control make it possible to equip robots, as well as traditional work machines, with wheel-specific motors and steering actuators. The increasing degree of freedom improves mobility and increases the complexity of the control. In addition to traditional Ackerman steering, the vehicle can be driven with skid or crab steering or a combination. On the other hand, the increased amount of DoFs can cause fatal errors if the control is not running properly. The speeds of the motors have to be in a certain relationship to each other and to the configuration of the steering actuators. On the other hand, the steering actuators always have to execute a realistic steering configuration. Additionally, the control system has to provide a safe and fluent transition from one driving state to another.

Multi-wheel control will be studied with simulation models of four-, six-, and eight-wheeled vehicles. A four-wheeled model with hydraulic actuators will be simulated by IHA and the six- and eight-wheeled vehicles by ATL. The four-wheeled model will be based on a hydraulic vehicle with articulated steering (Avant). The six-wheeled model is based on the Russian Marskhod, which has no steering actuators but additional actuators for inchworm-type rolking. The eight-wheeled model then has fully controllable wheel motors and a steering actuator for each wheel.

The control systems will first be built up in the simulator and tested with the models. Later the control systems will be coded and tested with the four-wheeled test vehicle (Avant), six-wheeled Marskhod, and eight-wheeled prototype vehicle of Patria.

Expected results: control systems for a four-, six-, and eight-wheeled vehicle with wheel-specific motors and steering, demonstration of the control with a real vehicle in the INTEGRATOR project

Task 2: Further development of rolking

The rolking locomotion system includes legs with the foot replaced by a wheel. In a rolking locomotion mode legs and wheels are used actively at the same time. Rolking, walking and rolling at the same time, has clear advantages on soft terrain where the conventional wheel does not work well. The need still exists to develop better control over rolking in order to allow fully automatic rolking and automatic change between wheeled locomotion and rolking. Automatic rolking will also be supported by developing laser- and camera-based methods of perceiving what lies ahead of the robot. This work will be carried out mainly in the perception research package and tested in this RP. Using the wheeled leg as a sensor, it is also possible to collect information about terrain. Typical locomotion parameters, such as rolling resistance and friction between the wheel and terrain, can be calculated using leg sensor information. Terrain geometric information could also be gained using the wheeled leg information. Terrain information can be used as a sensor input when an unknown area is mapped.

Expected results: control systems for a four-, six-, and eight-wheeled vehicle with wheel-specific motors and steering, demonstration of the control with a real vehicle in the INTEGRATOR project

Task 3: Further development of ball locomotion

Ball locomotion is based on a ball-shaped cover and an unbalanced rotating mass inside the ball. Propulsion can also be generated by an external force, such as wind or gravitation. ATL has already been researching locomotion for several years. Previous applications have been the Rollo home robot (diameters from 30-40 cm) and the tumbleweed locomotion mechanics (diameter 1.5 m) for ESTEC. In the home environment a ball shape looks friendly; it can easily be kicked away if it is under somebody’s feet. A camera and other optical perception and communication devices can easily be mounted under the transparent cover. In applications such as guarding and gas sensing the cover can be hermetically sealed. At the moment work on a society of small-sized ball-shaped guard robots is in progress.

Further development includes gaining a better understanding of ball dynamics in both a lowspeed mass (the existing systems) and in a gyro-type control where the unbalanced mass is rotating rapidly. This will be done by means of careful dynamic simulation. The simulated ball models will be completed with controllers tuned to the models. Simulated controllers will be programmed in C and implemented and tested with the real ball robot hardware.

Expected results: simulation of two different ball models and their controllers, implementation and testing of the controllers with real ball locomotion hardware

Task 4: Techniques for passive underwater locomotion

On the basis of the research carried out at ATL (SUBMAR) and on an active EU project (SWARM), there is a clear interest in continuing the study of passive underwater locomotion techniques. The word passive means here that there will be no active systems (i.e. propellers or thrusters) used for the locomotion. So far the study has been focused on Lagrangian drifter-type robots. These robots can only adjust their specific weight but are otherwise moved by the surrounding sea currents. In the future this research task will also study some other ways to move under water. One of the most promising research areas is the use of a glider-type robot for long-term and large-scale oceanographic research purposes. The glider is a drifter that, when adjusting its buoyancy, will, in a way, fly through the water columns. In that way it can cover quite large distances and stay operational for a long time, thanks to its low power consumption.

In order to produce coherent locomotion in a multi-robot case, novel locomotion algorithms must be developed. These algorithms will use the knowledge (a priori and estimated) that they have about the strengths and directions of the sea currents at different depths. The centre point of the swarm is calculated and a watchdog software module will sound the alert if some member of the swarm is getting too far from the centre point. If so, the unit will initiate locomotion procedures aimed at reducing the distance to an acceptable level. It is a highly complex research task with many constraints.

Expected results: algorithms for effective passive locomotion techniques for underwater conditions. They will be tested with simulators and with real robotic systems.

Task 5: Biologically inspired locomotion systems

Natural locomotion systems in the air, on the ground, and in water will be studied carefully together with biologists. Their capability and applicability for artificial mechanization will be analysed. The most promising systems (max. 5) will be studied more carefully by making preliminary performance calculations, kinematic and dynamic modelling, and finally an architectural manufacturing design with power and performance calculations. All the systems which seem to be feasible after this study will be implemented with full mechanics and control.

Expected results: new innovative motion system concepts