Addressing uneven flooring in 3D object manipulation tasks using collaborative mobile robotics & decentralised impedance control

Current methods of manipulating and transporting large objects within industry typically require infrastructure and bespoke fixtures and jigs. Traditional methods such as cranes are still used today but require a manual operator and commitment to a large space, and are then limited to that workspace to manipulate objects. More modern approaches use large mobile platforms such as the Kuka OmniMove platform to transport large objects to the desired location. An alternative to this approach is to use lots of smaller mobile robots which collaborate to transport the object to the target destination.

However, based on the literature reviewed in this area, all of the research undertaken in this field makes the large assumption that the floor the mobile robots are working on is perfectly flat and level. This is unrealistic, as the floors in many industrial settings are simply not flat nor level.

In this thesis, a novel approach to modelling collaborative mobile robots carrying an object together over uneven ground is proposed and explained. This approach takes inspiration from the automotive industry and considers the manipulators on the mobile platform as a ‘Suspension’ system. This is achieved by using an impedance control law, enabling the manipulators to act as mass spring dampers in multiple degrees of freedom. The modelling approach is generic, allowing the system to be scaled. In this work only four mobile robots are used for proof of concept and validation purposes.

Within the thesis, both ‘Passive’ & ‘Semi-active’ suspension are tested, the first using pure impedance control, whilst the second adapts the impedance controller to include a skyhook controller, which is a well-documented controller from the automotive industry.

Detailed modelling and experimental results are presented to support the proposed method. The experiments involve a team of robots traversing various floor disturbances commonly found in industry, which can lead to undesired forces being applied to the shared object. The focus of the results is on how the forces are distributed among the robots in the team. The experiments show that implementation of impedance control enables the team to distribute the forces more effectively, reducing stress on the object.

This is for both the ‘Passive’ & ‘Semi-Active’ systems. The performance of the controllers varies across different disturbances due to several factors. First the control parameters are not optimised for either of the controllers. Secondly, drift in the mobile platform itself, can cause issues to occur with the manipulators’ ability to compensate for the disturbances. Both of these limitations are further discussed in the relevant chapters, were the key areas of novelty are addressed.