Algorithmic robotics -- including but not limited to motion planning, task planning, manipulation, and mechanism design -- has been actively studied for more than three decades and has produced a rich collection of robotic algorithms that have the potential to make an intelligent robotic system that can achieve all industrial tasks automatically and efficiently. However, there is very limited use of these advanced algorithms on industrial robots, e.g., instead of automated planning, the teaching mode is still the dominant manner to set the robot’s trajectory for a given task. This workshop addresses this gap and the issues that arise in terms of developing appropriate planning and manipulation algorithms/methods that can be used on current and upcoming industrial platforms or enable technology transfer. These include addressing issues related to dynamic constraints, modeling uncertainty, computation of motion/grasp/manipulation strategies, development of software libraries, integration, and implementation, etc. The program consists of invited talks from leading academic researchers as well as industrial practitioners. Furthermore, we would schedule enough time for questions and discussions. This workshop will be of great interest to academics working in motion planning and manipulation, as well as the researchers and developers in the industry. We have a good mix of confirmed speakers from academia and industry, who would present different perspectives and issues.
Call For Contributions
    Participants are invited to submit abstracts related to key challenges while applying planning and manipulation techniques in real world industrial applications. Topics include but are not limited to: motion planning, task planning, manipulation, and mechanism design. We invite submissions in the form of extended abstracts (up to 4 pages) following RSS formatting guidelines. The abstracts will be reviewed by the organizers. Accepted contributions will be featured in a poster session and will be included in the workshop proceedings, which will be available at the workshop webpage. We encourage work-in-progress to be submitted and will take this into account in the review process.
Paper submission deadline: May 20, 2016. Submit via eMail to rss16irt.poster@gmail.com Submission Acceptance Notification: May 30, 2016.
Invited Speakers
  • Nagatani Tatsuya (Mitsubishi Electric)
    • Title: Robot Programming for Assembly Tasks
    • Abstract: We introduce some issues of robot programming in the production line for small and medium-sized electronics. And, we present two research results on it. 1st is calibration technology that is related to motion planning. This can be used to recovery robot systems, and to connect simulation and real system. 2nd is the grasp planning with small computation.
  • Dinesh Manocha (The University of North Carolina at Chapel Hill)
    • Title: Motion Planning for Industrial Robots
    • Abstract: Algorithmic motion planning has been actively studied in robotics and related areas for more than three decades. There is a rich collection of techniques that have been successfully used for CAD/CAM, bioinformatics, computer gaming and other applications. However, current techniques have two major limitations in terms of applications to physical robots. They are mostly designed for static environments or assume an exact geometric representation of all the obstacles in the scene. In this talk, we give a brief overview of our recent work on proximity queries and motion planning for high-DOF robots.We present novel techniques to perform collision and proximity queries with point-cloud sensor data,collected using depth or other sensors. We also present new planning algorithms based on optimization-formulation that take into account collision-free and dynamics constraints, and handles dynamic obstacles using a replanning framework. We demonstrate its application to human-like robots with dynamic constraints and Cartesian planning of industrial manipulators. They are also combined with sensor data and used for safe motion planning. We highlight their performance on human-like models with 26-DOF, PR2 robot with active sensing and a KUKA manipulator.
  • Philip Freeman (Boeing Research & Technology)
    • Title: Automated task and path planning in Aerospace Robotics
    • Abstract: Boeing is continuing to advance the use of robotics and automation in aerospace production. Boeing leverages automation to ensure high-quality products while reducing dull, dirty, and dangerous work. One active area of research is in maturing automated task and trajectory planning for advanced automation. These technologies have application to production processes as diverse as drilling and installing fasteners, painting aircraft, part measurement, and sealant application. In this presentation, I present some of the research challenges that Boeing sees in getting automated task and trajectory planning from practical academic solutions to industrial reality.
  • Andre Gaschler (Fortiss Institute, Technische Universität München)
    • Title: Task and motion planning for industrial robotics
    • Abstract: Because programming an industrial robot is a major effort, industrial applications are currently limited to large lot sizes and fixed paths. In this talk, we discuss methods for task-level programming, manipulation planning with semantic constraints, and integrated task and motion planning. The goal of this task-level approach is to allow easy robot programming without expert knowledge, flexible manufacturing methods, and automatic planning for variants and "lot size 1". For intuitive task-level programming, we demonstrate how CAD constraints are sufficient to implement practical assembly tasks. With all process information given on a semantic level, motion paths can automatically be solved for customized products. Finally, we integrate both task planning and collision-free motion planning in a hybrid search. This integrated planner can solve bi-manual assembly tasks generically for arbitrary initial placement of parts and robots while avoiding all types of collisions.
  • Michael Koval (CMU)
    • Title: Robot configuration estimation on high-dimensional implicit contact manifolds
    • Abstract: We investigate the problem of using contact sensors to estimate the configuration of a robot. Contact sensors are unique because they inherently discriminate between "contact" and "no-contact" configurations. The manifold particle filter (MPF) is a principled way of estimating state when state moves between manifolds of different dimensionality. Prior applications of the MPF were limited to two or three dimensions because of the difficulty of explicitly representing a high-dimensional contact manifold. In this work, we extend the MPF to higher dimensions by implicitly representing the contact manifold as the zero iso-contour of a signed distance function and sampling using constraint projection. We show that this formulation outperforms a conventional particle filter at estimating the configuration of an underactuated hand and two, three, and seven DOF arms.
  • Kensuke Harada (Osaka University)
    • Title: Planning Industrial Bin-Picking Tasks
    • Abstract: We will talk three topics on the motion planning for bin-picking tasks. We first explain the model-free bin-picking of objects where the shape of each object can be approximated by a cylinder. Then, we explain the motion planning of a dual-arm manipulator performing the industrial bin-picking task. Finally, we explain about the learning approach on bin-picking task.
  • Pieter Abbeel (UC Berkeley)
    • Title: Deep Reinforcement Learning for Robotics
    • Abstract: Deep learning has enabled significant advances in supervised learning problems such as speech recognition and visual recognition.Reinforcement learning provides only a weaker supervisory signal,posing additional challenges in the form of temporal credit assignment and exploration. Nevertheless, deep reinforcement learning has already enabled learning to play Atari games from raw pixels (without access to the underlying game state) and learning certain types of visuomotor manipulation primitives. I will discuss major challenges for, as well as some preliminary promising results towards, making deep reinforcement learning applicable to real robotic problems.
  • Eiichi Yoshida (AIST Japan)
    • Title: Human and humanoid motion planning for product design
    • Abstract: Human motion understanding and analysis is important for product design. Although have been developing digital human for product design, its focus has been to model the human shape. For effective human-centered design, human motion modeling and interaction with products or environment is an essential issue. We are currently developing an enhanced digital human model that can reproduce human motion based on measurement and simulate physical interactions, in order to help product design and to reduce the hardware prototyping. As we also believe that evaluation of real hardware, we have been investigating a humanoid robot that evaluates the products by actually using them like human subjects. The advantage of this method is repeatability, quantitative evaluation, and clearance of ethical issues. An ongoing work is shown on human motion reproduction by a humanoid for assistive device evaluation.
  • Levi Armstrong (Southwest Research Institute)
    • Title: Motion Planning with STOMP
    • Abstract: The vast majority of planners in MoveIt are randomized planners. These types of planners generate plans with large joint motions that appear unnatural from the observer’s perspective. This behavior is undesirable in an industrial application. The solution is to move away from randomized planners and towards optimization based planners. STOMP incorporates a smooth trajectory update rule forcing the trajectory to always be smooth in joint space. The smooth update rule is key to STOMP’s ability to generating sensible robot trajectories from an observer’s perspective. SwRI is focused on improving existing optimization based motion planners like STOMP (Stochastic Trajectory Optimization for Motion Planning), which provide an intelligent methodology for searching the joint space to achieve a collision free trajectory. SwRI will provide an overview of the ROS implementation of STOMP along with its advantages and disadvantages while providing an industrial application utilizing STOMP.

 Time  Title  Speaker

 Motion Planning for Industrial Robots  Dinesh Manocha (UNC Chapel Hill)
 Robot Programming for Assembly Tasks  Nagatani Tatsuya (Mitsubishi Electric)
   Coffee break
 Planning Industrial Bin-Picking Tasks  Kensuke Harada (Osaka University, Japan)
 Human and humanoid motion planning for product design  Eiichi Yoshida (AIST Japan)
 Task and motion planning for industrial robotics  Andre Gaschler (Fortiss Institute, TUM)
 12:25-12:40  Mobile Manipulation, collaboration and grasping with tactile sensing  Hao Zhang (Dorabot Inc.)
 12:40-02:30    Lunch Break
 02:30-03:05  Deep Reinforcement Learning for Robotics  Pieter Abbeel (UC Berkeley)
 03:05-03:40  Motion Planning with STOMP  Levi Armstrong (Southwest Research Institute)
 03:40-04:10    Coffee Break
 04:10-04:45  Automated task and path planning in Aerospace Robotics  Philip Freeman (Boeing Research & Technology)
 04:45-05:20  Robot configuration estimation on high-dimensional implicit contact manifolds  Michael Kovel (CMU)
 05:20-05:35  Using Object-Oriented Development for Planning and Controlling Industrial Robot Systems  Andreas Schierl (University of Augsburg)
 05:35-05:50  Fully sample-based configuration free space approximation for optimal motion planning  Donghyuk Kim (KAIST)
 05:50-05:55    Closing
Intended Audience
Our intended audience would include researchers and developers who are working in planning, manipulation, mechanism and are interested in solving real challenges in the industry using the state-of-the-art robotics techniques developed in the academic community. This workshop will demonstrate the progress in the robot-related industry (e.g., service robots, manufacturing, logistics), introduce main problems and challenges in real industry, and finally discuss the gap and its possible solutions while applying research outputs to the real industry.
There have been several workshops related to planning techniques for industrial robots at ICRA and IROS over the last few years. We have co-organized successful workshops on “Planning and Programming Technologies for Industrial Robots”, in 2013 IREX (Tokyo), “Motion Planning for Industrial Robots” at ICRA 2014, and “Caging and Its Applications in Grasping/Multiagent Cooperation” at IROS 2013; Norbert Krüger et al. organized the “Transfer of Cognitive Robotics Research to Industrial Assembly and Service Robots” at IROS 2015.