Executive Summary : | This proposal aims to address the cooperative manipulation and transportation of an object, in the presence of static obstacles, using mobile robots. Successfully moving an object from its start pose to its goal pose may require repeated orientation changes to the object (referred to in this proposal as manipulation), followed by transportation. Even though this is a research problem in the field of multi-robot systems, it is a common activity in our daily life and one can also find examples from nature (ants carrying food). Over the recent years, robotic systems have been deployed extensively for manipulation and transportation tasks in warehouses. But majority of these applications involve a single robot interacting with an object in a given instant. This in turn limits the size and weight of the object involved in the task. The solution would be to use multi-robot cooperative manipulation and transportation strategy. While cooperative manipulation and transportation of objects has been explored before in the research community, this technology still remains limited to research applications. Direct application of motion planning techniques is not feasible for these systems due to their high dimensionality. Moreover, multi-robot applications will have additional constraints to account for the robot-object interactions. The proposed research aims to address the challenges of cooperative manipulation and transportation through a novel multistage approach. The high-level planning stage will compute a feasible path for the object from start to goal while avoiding the obstacles. The low-level Model Predictive Controller (MPC) computes the control commands to the individual robots to successfully manipulate and transport the object along the planned path. Unlike existing works in this domain, this approach will take into consideration the coupled dynamics of the robot-object system through the multi-body dynamics model used by the MP controller. The developed cooperative manipulation and transportation framework will be evaluated over varying scenarios in both simulated and real-world conditions. To this extent, the proposed research will also include design and integration of two differential drive robots by leveraging open-source designs, hardware, and software while ensuring that the overall cost remains low. After sufficient evaluation, the framework will be released as an open-source software package to allow for easy adaptation by the worldwide robotics research community. Developing cooperative manipulation capabilities for multirobot systems will be the next milestone towards bringing autonomous systems closer to real-world applications. With this target, the final objective of this research effort will be to deliver a viable solution that can aid in a variety of real-world applications such as material handling in construction, factory, and warehouse environments. |