[robotics-worldwide] [jobs] Research Internship Opening on "3D Perception and Learning for Humanoid Robot Locomotion/Manipulation" at the Italian Institute of Technology (IIT), Genoa
Dimitrios.Kanoulas at iit.it
Thu Apr 12 01:47:16 PDT 2018
Fellow Junior (Research Internship) Opening on "3D Perception and Learning for Humanoid Robot Locomotion/Manipulation" at the Italian Institute of Technology (IIT), Genoa
Where: Humanoids and Human-Centred Mechatronics Lab (HHCM), Department of Advanced Robotics (ADVR) at the Istituto Italiano di Tecnologia (IIT), Genoa, Italy.
Starting Date: As soon as possible; June-July 2018 (preferable)
Deadline: April 23, 2018 (a FCFS policy will be followed in the evaluation of the applications until a suitable candidate is appointed)
Duration: 4-5 months with potential follow up enrollment to continue for PhD studies.
Potential Topics (one should be selected):
-- "3D Perception and Learning for Legged Robot Navigation: rough terrain locomotion (vision and path planning) using geometric and machine learning techniques"
-- "3D Perception and (Deep) Learning for Humanoid Robot Tool Manipulation: detection, localization, and use"
Tutors: Dr. Dimitrios Kanoulas, Dr. Nikos Tsagarakis
Contacts: Dimitrios.Kanoulas at iit.it, nikos.tsagarakis at iit.it
The Humanoids and Human Centred Mechatronics Lab at the Department of Advanced Robotics of the Italian Institute of Technology (IIT) (an English language Institute, placed in Genoa) is seeking a fully funded 4-5 months research intern (Msc should have been completed), with a strong ambition to pursue a research career, to work on one of our EU-funded WALK-MAN (walk-man.eu, video: youtu.be/kZzwVwzAWME, video: youtu.be/7UectbjusCI), CogIMon (cogimon.eu), and/or CENTAURO (centauro-project.eu) humanoid robots.
After Fukushima Daiichi nuclear disaster in 2011 the need of robots to deal with unstructured environments and replace humans in hazardous tasks became one of the main open problems in robotics. Rapid advancements in actuation and control over the last few years enabled articulated legged robots to walk in uneven terrain and manipulate several objects. Still, the problems of 1) foot placement in rough terrain for navigation and 2) tool manipulation and reasoning in uncertain setups using 3D perception, remain two of the main challenges in robotics and are the key aspects for completing several tasks in unknown environments.
The aim of this topic is to develop new geometric or machine learning (e.g. deep learning) methods for environment reconstruction and cognition that enable limbed robots, such as WALK-MAN (walk-man.eu), CENTAURO (centauro-project.eu) , or CogIMon (cogimon.eu) to perform legged locomotion and manipulation in unstructured environments using 3D perception for foot-placement and hand/arm/wholebody-placement for path planning and mapping (SLAM). Several exteroceptive (stereo/event/rgb cameras, RGB-D sensors, 2D/3D lidar scanners) sensors will be used on our limbed robots, to acquire RGB images and dense 3D point clouds. Geometric simplifications for reasoning the contact between the robot's body (e.g., foot, hands, arms) and the environment will need to take place. Moreover, path planning methods need to be developed to extract primitives for locomotion and manipulation. The development and testing will take place on our full-size humanoid/quadrupedal robots in real-world environments.
1. To be eligible, you must have recently (at most 6 months) completed a Masters degree, or expect to complete soon.
2. This topic lies in the intersection of Vision, Learning, and Robotics. A background in any of Robotics, Computer/Robotic Vision, Path Planning, Robot Learning is desirable.
3. Ideal applicants should have strong C++ (Python and Matlab are a plus) programming skills.
4. Machine learning and computer vision skills are required.
5. Knowledge of the Robot Operating System (ROS) and the Point Cloud Library (PCL) is a very big plus.
6. The applicants should be fluent in English and team players.
-- Legged Robot Navigation
1. Kanoulas et. al., "Vision-Based Foothold Contact Reasoning using Curved Surface Patches", Humanoids 2017.
2. Kanoulas et. al., "Footstep Planning in Rough Terrain for Bipedal Robots using Curved Contact Patches", ICRA 2018.
-- Humanoid Robot Tool Manipulation
1. Kanoulas et. al. "Center-of-Mass-Based Grasp Pose Adaptation Using 3D Range and Force/Torque Sensing", IJHR 2018.
2. Nguyen et. al., "Object-Based Affordances Detection with Convolutional Neural Networks and Dense Conditional Random Fields", IROS 2017.
The applicants should submit their CV electronically outlining experience and qualifications, a short statement (max 1 page) about the research areas of interest as well as the names of two referees to Dimitrios.Kanoulas at iit (cc-ed nikos.tsagarakis at iit.it), quoting [HHCM-INTERNSHIP] in the email subject (total email size should not exceed 1MB).
Best Regards --
Dimitrios Kanoulas, Ph.D.
Humanoids and Human-Centred Mechatronics Lab (HHCM)
Dept. of Advanced Robotics (ADVR)
Istituto Italiano di Tecnologia (IIT-Genova)
Via Morego 30,Genova 16163, Italy
e-mail: Dimitrios.Kanoulas at iit.it
More information about the robotics-worldwide