Master's Thesis
Learning Mobile Robot's Push Action Models for Box Transportation in Logistics Application
— 2022
Key information
Authors:
Supervisors:
Published in
November 30, 2022
Abstract
Robot automation can allow companies to stay competitive in a world where there is a push to incorporate automation into the production factors of companies. As such, the usage of robots in companies' warehouses to automate their logistics is a step towards that direction. Furthermore, the use of non-prehensile manipulation may allow robots to perform better the task of handling objects in these warehouses. In this work, we want to learn mobile robot push action models. To that end, we use real data to train specialized prediction models. Furthermore, we also want to solve a pushing problem resembling a logistic task in a warehouse, where a mobile robot needs to push a box-shaped object to a desired pose. To that end, we propose a push planning algorithm that incorporates the learned forward prediction models in its planning. Our experimental results show that the learned models only need a small number of pushes per action to achieve reasonable prediction accuracy. Furthermore, these models have some generalization capability. We also show that these models can be used to solve our pushing problem. Finally, we proposed that, instead of using a single general model to generate a prediction, we select the model that best fits an object currently being pushed from a library of specialized models. Following this approach, we propose a model selection method called Greedy Model Selection.
Publication details
Authors in the community:
Supervisors of this institution:
Fields of Science and Technology (FOS)
electrical-engineering-electronic-engineering-information-engineering - Electrical engineering, electronic engineering, information engineering
Publication language (ISO code)
eng - English
Rights type:
Embargo lifted
Date available:
September 30, 2023
Institution name
Instituto Superior Técnico