Abstract
Programming robots to perform complex tasks is a very expensive job. Traditional path planning and control are able to generate point to point collision free trajectories, but when the tasks to be performed are complex, traditional planning and control become complex tasks. This study focused on robotic operations in logistics, specifically, on picking objects in unstructured areas using a mobile manipulator configuration. The mobile manipulator has to be able to place its base in a correct place so the arm is able to plan a trajectory up to an object in a table. A deep reinforcement learning (DRL) approach was selected to solve this type of complex control tasks. Using the arm planner’s feedback, a controller for the robot base is learned, which guides the platform to such a place where the arm is able to plan a trajectory up to the object. In addition the performance of two DRL algorithms ((Deep Deterministic Policy Gradient (DDPG)) and (Proximal Policy Optimisation (PPO)) is compared within the context of a concrete robotic task.
Highlights
Logistics applications demand the development of flexible, safe and dependable robotic solutions for part-handling including efficient pick-and-place solutions.Pick and place are basic operations in most robotic applications, whether in industrial setups or in service robotics domains
All the environments we modelled were integrated with Gym-gazebo, which enabled us to straightforwardly use OpenAI Baselines deep reinforcement learning (DRL) library, which is designed to work with Gym
500 evaluation periods were made uniformly distributed over the learning process and the metrics used to evaluate the performance were the mean accumulated reward and the success rate, which are the most used ones in the DRL community
Summary
Logistics applications demand the development of flexible, safe and dependable robotic solutions for part-handling including efficient pick-and-place solutions. That is not the case when it comes to manipulating parts with high variability or in less structured environments In this case, picking systems only exist at laboratory level, and have not reached the market due to factors such as lack of efficiency, robustness and flexibility of currently available manipulation and perception technologies. This study focused on the development of adaptive trajectory planning strategies for pick and place operations using mobile manipulators. This study focused on learning to navigate to such a place that the mobile manipulator’s arm is able to pick an object from a table. The simulated robot we used is the mobile manipulator miiwa, depicted in Figure 1, which has been used in industrial mobile manipulation tasks (e.g., [3])
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
