Biography

http://secure-robots.eu/Bruno Brito has obtained as Master Degree in Electrical Engineering and Computer Science from the University of Porto (Portugal) in 2013. During his master he has obtained several awards, such as the ABB and Continental prize for the best master thesis.

He gave is first steps outside of the university as trainee at ESA in the Guidance, Navigation and Control (GNC) section for 2 years where start researching about Space Robotics for space debris removal and Hardware-In-the-Loop simulations (http://www.esa.int/spaceinvideos/Videos/2015/10/AIM_mission_navigation_camera_testing).

Currently he is a researcher engineer and scientist in the Department “Robot and Assistive Systems” at Fraunhofer IPA since 2016. His research is mainly focused in safe manipulation and safe human-robot interaction.

Detailed CV.

Objectives

Nowadays, mobile robots are able to safely navigate in the most harsh environments. However their manipulation capacities are still limited when it comes to safe interaction with humans and the environment. The proposed method allow the robot to have passive and dynamic reactions, allowing it to safely interact in a human environment. The combination of passive control techniques with dynamical systems has been shown to be able to interact with the static and dynamic environments with fixed base manipulators while keeping a stable behavior. In this work we propose to extend these methods for mobile manipulators, creating new capabilities for Care-O-bot 4.

Results

Motion Planning

Several search-based and optimisation based motion planners were tested in simulation and in real robots (Care-O-Bot 4, rob@work) in order to evaluate their performance and drawbacks. The planners were tested in a pick and place scenario, in which the robot as to pickup several objects with different poses from a shelf similar to the AMAZON picking challenge. The results were presented at Stuttgart Vision Fair 2016.

Self-collision avoidance NMPC for Mobile Manipulators

Mobile robots are able to safely navigate in the harsh environments. However their manipulation capacities are still limited when it comes to safe interaction. Various methods tried to overcome some of the problems, such as:

• Unified Weighted Least Norm (UWLN)
• Gradient Projection (GPM)
• Stack of tasks

The first can not avoid self-collisions. The others are not collision free and can become unstable. Our control scheme consist on a finite horizon Nonlinear Model Predictive Controller (NMPC) allowing the robot to track a Cartesian trajectory while dealing with all the constraints. The predictive model allows the robot to anticipate and react faster to the imposed constraints. The results where presented in http://soma-summerschool.dlr.de/

Contacts

Email: bruno.brito@ipa.fraunhofer.de

Linkedin: https://www.linkedin.com/in/bruno-brito-1b815338/

Github: https://github.com/ipa-bfb