Willow Garage Blog
A Mobile Manipulation Challenge, organized by Bill Smart and Willow Garage's Matei Ciocarlie, will take place at ICRA 2010 in Anchorage, Alaska. The challenge aims to bring together a variety of mobile manipulation platforms, and provide the community the opportunity to investigate the current state of the art of mobile manipulation.
The event will draw together the mobile manipulation community to identify collaboration opportunities, recognize current accomplishments, and establish future research milestones. As such, the event is set up as a challenge and not a competition. There are no winners or losers, just participants eager to share their work with the community. The feasibility of performing mobile manipulation demos outside of the lab is often hindered by transport costs, but the impact of making these platforms visible to other researchers is profound. Therefore, some funding will be available to mitigate these expenses.
The deadline for submitting your Letter of Participation is December 1, 2009! Please click here for more information, and we look forward to seeing your work!
Nate Koenig of the Interaction Lab at USC is continuing his work here at Willow Garage after a busy summer. Nate carried out an empirical study investigating the use of people as teachers for robots, while also researching learning by demonstration with PR2.
Participants in Nate's study used learning by demonstration to teach PR2 how to solve the Towers of Hanoi puzzle. As the name implies, learning by demonstration relies on a human teacher to provide a robot "student" with demonstrations of a complex task. In this case, the robot uses the state of the puzzle (i.e., location of red disk compared to blue and green), along with the teacher's command, to learn the demonstrated task. Volunteers used a web-based teaching tool to guide PR2 through the three-disk puzzle board. In one condition, teachers were able to directly see PR2, while in the other condition teachers viewed the robot's actions through a small video feed on the web tool. This manipulation allowed Nate to study if robot visibility affects teaching strategies and outcomes. Based on participants' commands and other observations from the environment, the robot learned how to solve the puzzle on its own.
Results from this study indicate that teachers perform better when visually separated from the robot. Performing "better" means that participants made fewer unnecessary or repetitive moves when teaching the robot. While this may seem counter-intuitive, the teachers who could see the robot were easily distracted from the task and seemed to build inaccurate mental models of PR2's capabilities.
In addition to this experiment, Nate worked on a number of smaller projects including developing the first video streaming ROS node and creating a web-based graphical interface for interacting with PR2. You can find many of these contributions in the hanoi package for ROS. Nate is also creator and lead developer for Gazebo, a popular open-source 3D robot simulator. Gazebo is heavily used at Willow Garage to simulate the actions of PR2, and Nate is providing us with numerous improvements.
Here at Willow Garage, we're accustomed to robots roaming the hallways. Robots are our tools (and occasionally, entertainment), and we always know what's coming when we hear the familiar drone of casters down the hall. Recently, however, we can't be so sure. The newest addition to our building sports a single PR2 caster and resembles a scrawny metal stick figure -- until you notice the smiling, human face gazing back at you from the mounted computer monitor. Usually it greets you by name. This new addition is named Texas, and it's driven by Dallas. In Indiana.
Dallas Goecker is a Willow Garage electrical engineer living in Indiana, and Texas is Dallas' new telepresence robot here in the office. Dallas is Willow Garage's first telecommuter and it's only fitting that he telecommutes by robot. Dallas teleoperates his surrogate around the entire building, attending meetings, asking questions, working collaboratively, and, sometimes, taking lunch breaks with us. Like PR2, Texas sometimes loses wireless connectivity, but ultimately, Texas is a co-worker, not just a tool. We make small talk when passing in the hall, ask work-related questions, and politely offer to share our food, albeit, sarcastically. Through this telepresence robot, Dallas has been able to integrate himself with the rest of the company, and really become a member of the team despite his location.
Dallas, along with Curt Meyers, built the first prototype of Texas by raiding the spare-parts drawer at Willow Garage. They used an old prototype PR2 caster, a car battery, a leftover monitor, and some Bosch framing to build the basic structure. They added off-the-shelf speakers, microphone, web camera, and a laser range finder. They were able to use the same ROS software that we're using to build the PR2: motor controllers, navigation stack, and teleoperation software. All they had to change was the "robot model", which describes the basic structure of the robot.
Texas may not have all the bells and whistles of the PR2, but it shows us the new avenues and potential of our modular hardware and software. It may be built out of leftovers, but its simple capabilities have profoundly changed how we are able to interact with Dallas.
PR2 fabrication is progressing nicely, and the production team has built the first set of grippers. As the grippers are assembled, they undergo various tests to ensure high quality. For example, the initial qualification of a motor/encoder assembly tests motor speed and verifies the gear box's gear ratio. After assembly is complete, the grippers undergo a series of controlled drop tests to ensure that all hardware and electrical connections are secure. Next, the grippers are exercised, "burned-in", for 48 hours. During the burn-in, the gripper squeezes a hard rubber ball, and we monitor the amount of current drawn by the motor, as well as the relative positioning of the gripper fingers. The purpose of the burn-in is to weed out potentially defective components, and to verify that the screws holding the assembly together do not work loose. Only after completing all of these tests satisfactorily are the grippers allowed into the stockroom, where they wait to be issued to the next higher assembly of the PR2.
Kevin Watts talks about drop-testing and the burn-in process:
Caster assembly training commenced last week, and the first
set of sub-assembly kits are now on the floor. The engineering team is
doing Engineering Fit Verification (EFV) tests of the next assemblies to ensure that the parts meet specification. Once the kits are complete and the assembly documentation is
ready, base/body/spine assembly training will begin.
The stockroom is a busy place now as we continue to receive parts and issue kits to the production floor. We have 75% of the PR2 parts in the stockroom now, with final parts anticipated within the next 3 weeks.
-- John Blazek
For a little background: Wired: Company Denies its Robots Feed on the Dead.
See more of Jorge Cham's work at PhD Comics.
This work is licensed under a Creative Commons Attribution-Noncommercial 3.0 United States License.
ROS 0.9 has been released! Our focus with this release was bringing ROS closer to 1.0 status. ROS has been undergoing user testing as part of our Milestone 3 efforts, which has allowed us to make improvements to numerous tools, clarify error messages, and fix bugs. The most notable change with this release is that we have officially ended support for the deprecated ros::Node API in roscpp. We have also introduced a new "bool" type for ROS message types.
There were many other changes with this release as we work to improve and finalize the ROS feature set. You can find a complete list in the changelist.
We'll be very busy at IROS 2009. For those of you interested in Willow Garage-related talks, we've included a schedule below.
Monday, October 12
09:25-10:45, Video Sessions, Sterling 9
Stanford Testbed of Autonomous Rotorcraft for Multi-Agent Control: Gabriel Hoffmann (PARC), Steven Lake Waslander (University of Waterloo), Michael Vitus (Stanford), Haomiao Huang (Stanford), Jeremy Gillula (Stanford), Vijay Pradeep (Willow Garage), Claire Tomlin (UC Berkeley)
Abstract: The Stanford Testbed of Autonomous Rotorcraft for Multi-Agent Control, a fleet of quadrotor helicopters, has been developed as a testbed for novel algorithms that enable autonomous operation of aerial vehicles. The testbed has been used to validate multiple algorithms such as reactive collision avoidance, collision avoidance through Nash Bargaining, path planning, cooperative search and aggressive maneuvering. This article briefly describes the algorithms presented and provides references for a more in-depth formulation, and the accompanying movie shows the demonstration of the algorithms on the testbed.
14:00 - 15:40, Mapping I, Grand C
Abstract: The typical SLAM mapping system assumes a static environment and constructs a map that is then used without regard for ongoing changes. Most SLAM systems, such as FastSLAM, also require a single connected run to create a map. In this paper we present a system of visual mapping, using only input from a stereo camera, that continually updates an optimized metric map in large indoor spaces with movable objects: people, furniture, partitions, etc. The system can be stopped and restarted at arbitrary disconnected points, is robust to occlusion and localization failures, and efficiently maintains alternative views of a dynamic environment. It operates completely online at a 30 Hz frame rate.
Tuesday, October 13
10:50-11:10, Smart Actuators, Mills 7
Gunter Niemeyer will be co-chairing the Smart Actuators session.
Variable Impedance Magnetorheological Clutch Actuator and Telerobotic Implementation: Daniel Walker (Stanford), Dan Thoma (Los Alamos National Laboratory), Gunter Niemeyer (Willow Garage).
Abstract: Variable impedance actuation is characterized by the ability to independently set output force and output impedance for a robotic device. Adjusting the output impedance in real-time allows a device to better adapt to a variety of tasks, operate in human-like fashion, and support human safety. This paper focuses on a Series Clutch Actuator based on magnetorheological fluid which allows a fast, electrical change of the impedance while maintaining good force tracking. In particular the mechanical clutch can alter the high-frequency impedance, decoupling the motor inertia and thus reducing impact forces. We present the mechanical clutch design and a control system architecture to automatically adjust the fluid magnetization level and leverage the clutch benefits. Experiments verify torque tracking and impact force reduction both in autonomous and telerobotic operation. The actuator was designed and manufactured in collaboration with the Materials Design Institute at Los Alamos National Laboratory, and is tested in a single degree of freedom demonstration.
17:20-17:40, Haptics IV, Grand H
Gunter Niemeyer will be co-chairing the Haptics IV session.
Improved Multi-DOF Haptics with Spring Drive Amplifiers: Robert Wilson (Stanford), Gunter Niemeyer (Willow Garage)
Abstract: Spring drive amplifiers utilize the natural inductive properties of a DC motor to produce stronger position feedback than is achievable with traditional current amplifiers. Impedance-type haptics devices can leverage these amplifiers to render superior contacts with higher contact stiffness. Use of these amplifiers, however, requires the virtual environment to specify motion commands for each motor. This work extends the recently developed 1-DOF integrated approach to multi-DOF applications. It presents a motion controller working with spring drive amplifiers, together producing the maximum achievable isotropic Cartesian stiffness. Isotropy is necessary to guarantee that haptic contact forces are rendered in the correct direction. The system is implemented on a PHANTOM 1.0 haptic device and analytic performance is verified by experiment.
Wednesday, October 14
9:30-9:50, Personal Robots, Mills 3
Real-Time Perception-Guided Motion Planning for a Personal Robot: Radu Rusu (TUM), Ioan Sucan (Rice), Brian Gerkey (Willow Garage), Sachin Chitta (Willow Garage), Michael Beetz (TUM), Lydia Kavraki (Rice).
Abstract: This paper presents significant steps towards the online integration of 3D perception and manipulation for personal robotics applications. We propose a modular and distributed architecture, which seamlessly integrates the creation of 3D maps for collision detection and semantic annotations, with a real-time motion replanning framework. To validate our system, we present results obtained during a comprehensive mobile manipulation scenario, which includes the fusion of the above components with a higher level executive.
14:40-15:00, Telerobotics - Haptics, Grand H
Gunter Niemeyer will be co-chairing the Telerobotics - Haptics session.
Open-Loop Bilateral Teleoperation for Stable Force Tracking: Peter Shull (Stanford), Gunter Niemeyer (Willow Garage)
Abstract: Traditional bilateral teleoperation communicates both motion and force information explicitly between master and slave devices. Any such closed loop architecture trades off performance with potential instability, especially when using force measurements of high inertia slaves contacting stiff environments. More conservatively, open-loop architectures avoid stability issues, transmitting motion commands while allowing any force feedback only via sensory substitution. We propose open-loop bilateral teleoperation as an alternative communicating force information explicitly and restricting motion information to visual feedback. This naturally matches a user's needs, seeing motion and feeling forces. A user study was conducted to compare the novel user interface to three common open loop and bilateral control methods: position control, position control with force feedback, and rate control. The results of this study show that users are able to achieve superior force tracking with little tremor. Position tracking and trial completion time suffered from the lack of direct position connection, but training provides a promising method to restore this performance.
16:20-16:40, Human Robot Interaction VI, Grand B
Abstract: As robots enter the everyday physical world of people, it is important that they abide by society's unspoken social rules such as respecting people's personal spaces. In this paper, we explore issues related to human personal space around robots, beginning with a review of the existing literature in human-robot interaction regarding the dimensions of people, robots, and contexts that influence human-robot interactions. We then present several research hypotheses which we tested in a controlled experiment (N=30). Using a 2 (robotics experience vs. none: between-participants) x 2 (robot head oriented toward a participant's face vs. legs: within-participants) mixed design experiment, we explored the factors that influence proxemic behavior around robots in several situations: (1) people approaching a robot, (2) people being approached by an autonomously moving robot, and (3) people being approached by a teleoperated robot. We found that personal experience with pets and robots decreases a person's personal space around robots. In addition, when the robot's head is oriented toward the person's face, it increases the minimum comfortable distance for women, but decreases the minimum comfortable distance for men. We also found that the personality trait of agreeableness decreases personal spaces when people approach robots, while the personality trait of neuroticism and having negative attitudes toward robots increase personal spaces when robots approach people. These results have implications for both human-robot interaction theory and design.
08:30-17:30, Mills 9, Semantic Perception for Mobile Manipulation
This work is licensed under a Creative Commons Attribution-Noncommercial 3.0 United States License.
Ben Cohen of University of Pennsylvania has returned to the GRASP Lab after his summer internship here at Willow Garage. At Penn, Ben researches search-based methods for path planning for robotic manipulators. During his time here, he worked on two motion planners: one for door opening and one for manipulation. When compared to the door planner used in Milestone 2, the new door planner uses SBPL (Search-Based Planning Library) to give the PR2 two new capabilities. First, it allows the robot to not only push doors open, but also pull. Second, the door planner allows the robot to open doors, regardless of hinge position -- left or right side of the door. These two novel capabilities allow for robust, more universal door opening.
Additionally, Ben's work on a manipulation planner involved integrating SBPL into the move_arm ROS package, which integrates a variety of motion planners. Ben tested the SBPL planner on the PR2's arms, and added the supporting software needed to perform collision checking. With collision checking in place, SBPL can more readily handle cluttered, complex environments.
Here are Ben's end-of-summer presentation slides discussing his planning work (Download PDF from ROS.org):
Conor McGann and Bhaskara Marthi went to ICAPS 2009 to give presentations on planning research they are working on at Willow Garage. This includes some of the work that went into Milestone 2 and TREX, as well as more on the hierarchical planning work seen in Jason Wolfe's summer internship work. For those that couldn't make it to Greece or want a second look, their presentation slides are included below.
Conor McGann's presentation on "Model-based Hierarchical Control of a Mobile Manipulation Platform", which was given at the workshop on Planning and Plan Execution for Real-World Systems (Download PDF from ROS.org):
Bhaskara Marthi's presentation on Angelic Hierarchical Planning, which was given at the workshop on Bridging the Gap Between Task and Motion Planning (Download PDF from ROS.org):