Embedded Robotics: Mobile Robot Design and Applications with Embedded Systems

Decentralized reinforcement learning of robot behaviors

In grasping tasks carried out with humanoids, knowledge about the robot's reachable workspace is important Without this knowledge, it might be necessary to repeatedly adapt the stance location and call an inverse kinematics solver before a valid robot configuration to reach a given grasping pose can be found In this paper, we present an approach to select an optimal stance location in SE(2) for a humanoid robot's feet relative to a desired grasp pose We use a precomputed representation of the robot's reachable workspace that stores quality information in addition to spatial data By inverting this representation we obtain a so-called inverse reachability map (IRM) containing a collection of potential stance poses for the robot The generated IRM can subsequently be used to select a statically stable, collision-free stance configuration to reach a given grasping target We evaluated our approach with a Nao humanoid in simulation and in experiments with the real robot As the experiments show, using our approach optimal stance poses can easily be obtained Furthermore, the IRM leads to a substantially increased success rate of reaching grasping poses compared to other meaningful foot placements within the vicinity of the desired grasp

Stance selection for humanoid grasping tasks by inverse reachability maps

In this paper, we present a walking approach for the Nao robot that improves the agility and stability of the robot when walking on a flat surface such as the soccer field used in the Standard Platform League. The gait uses the computationally inexpensive model of an inverted pendulum to generate a target trajectory for the center of mass of the robot. This trajectory is adapted using the observed real motion of the center of mass. This approach does not only allow compensating the inaccuracies in the model, but it also allows for reacting to external perturbations effectively. In addition, the method aims at facilitating a preferably fast walk while reducing the load on the joints.

/pdf/a-center-of-mass-observing-3d-lipm-gait-for-the-robocup-3ghszug7wm.pdf

A center of mass observing 3D-LIPM gait for the RoboCup standard platform league humanoid

Before beginning any robot task, users must position the robot's base, a task that now depends entirely on user intuition. While slight perturbation is tolerable for robots with moveable bases, correcting the problem is imperative for fixed- base robots if some essential task sections are out of reach. For mobile manipulation robots, it is necessary to decide on a specific base position before beginning manipulation tasks. This paper presents Reuleaux, an open source library for robot reachability analyses and base placement. It reduces the amount of extra repositioning and removes the manual work of identifying potential base locations. Based on the reachability map, base placement locations of a whole robot or only the arm can be efficiently determined. This can be applied to both statically mounted robots, where the position of the robot and workpiece ensure the maximum amount of work performed, and to mobile robots, where the maximum amount of workable area can be reached. The methods were tested on different robots of different specifications and evaluated for tasks in simulation and real world environment. Evaluation results indicate that Reuleaux had significantly improved performance than prior existing methods in terms of time-efficiency and range of applicability.

Reuleaux: Robot Base Placement by Reachability Analysis

Complex motions like kicking a ball into the goal are becoming more important in RoboCup leagues such as the Standard Platform League. Thus, there is a need for motion sequences that can be parameterized and changed dynamically. This paper presents a motion engine that translates motions into joint angles by using trajectories. These motions are defined as a set of Bezier curves that can be changed online to allow adjusting, for example, a kicking motion precisely to the actual position of the ball. During the execution, motions are stabilized by the combination of center of mass balancing and a gyro feedback-based closed-loop PID controller.

/pdf/kicking-a-ball-modeling-complex-dynamic-motions-for-humanoid-3tiday1sde.pdf

Kicking a ball - modeling complex dynamic motions for humanoid robots

In this paper we introduce an online grasping system for the Nao robot [1] manufactured by Aldebaran Robotics. The proposed system consists of an object detector and a grasp motion planner. Thereby, known objects are detected by a stereo contour-based object detector and hand motion paths are planned by an A*-based algorithm while avoiding obstacles. Compared to skilled robots such as Justin [2] or ASIMO [3] online grasping with the Nao constitute as particular problem due to the limited processing power and the hand design. The methods proposed allow detecting and grasping objects in real-time on an affordable humanoid robot.

/pdf/grab-a-mug-object-detection-and-grasp-motion-planning-with-40vl5w3w3a.pdf

Grab a mug - Object detection and grasp motion planning with the Nao robot

The aim of a soccer game is to score more goals than the opponent does. The chances of doing so increase when all the own players are continuously present on the field. In the RoboCup Standard Platform League (SPL), the only reasons for players not to participate in a game are that they either have been penalized or they have not been ready for the match in the first place. This paper presents some of the methods the 2013 SPL world champion team B-Human uses to ensure a stable and continuous game performance. These include overcoming the weak obstacle avoidance of our 2012 system, reacting to a rule change by an improved detection of the field boundary, better support for analyzing games by logging images in real time, and quality management through procedures the human team members follow when participating in a competition.

/pdf/b-human-2013-ensuring-stable-game-performance-1knsov5mfg.pdf

B-Human 2013: Ensuring Stable Game Performance

In this paper we introduce an online object manipulation system for the NAO robot that is able to detect and grasp an object out of a human hand and then give it back in real-time. Known objects are rendered from 3D models and detected stereo contour-based by using a new stereo vision head for NAO. In order to grasp objects, motion trajectories are generated by an A* planner while avoiding obstacles. In order to safely release objects back into a human hand, a combination of tactile and force sensors of the carrying arm is used to detect whether someone touched the grasped object. We performed quantitative experiments in order to evaluate the quality of the detector, the time to grasp an object, as well as the number of successful grasps. We demonstrated the whole system on the real robot.

GRASPY – Object Manipulation with NAO

B-Human is a joint RoboCup team of the Universitat Bremen and the German Research Center for Artificial Intelligence (DFKI). The team consists of numerous students as well as three researchers. The latter have already been active in a number of RoboCup teams such as the GermanTeam and the Bremen Byters (both Four-Legged League), B-Human and the BreDoBrothers (Humanoid KidSize League), and B-Smart (Small-Size League). We entered the Standard Platform League already in 2008 as part of the BreDoBrothers, a joint team of the Universitat Bremen and the Technische Universitat Dortmund, providing the software framework, state estimation modules, and the get up and kick motions for the NAO. For RoboCup 2009, we discontinued our Humanoid Kid-Size League activities and shifted all resources to the SPL, starting as a single location team after the split-up of the BreDoBrothers. Since the split-up, the team B-Human has won every tournament it participated in. In 2009, 2010, and 2011, we won the RoboCup as well as the RoboCup German Open. This year, we already won the RoboCup German Open in Magdeburg, and we hope to be able to continue our success by also winning the RoboCup in Mexico City. This team description paper is organized as follows: Section 2 presents all current team members, followed by short descriptions of our publications since the last RoboCup in Sect. 3. Our major development, the ability to play on a field without unique cues, is described in Sect. 4. This is accompanied by several developments regarding walking and proprioceptive sensing in Sect. 5 as well as multiple changes to our infrastructure in Sect. 6.

Judith Müller

Papers

Kicking a ball - modeling complex dynamic motions for humanoid robots

Grab a mug - Object detection and grasp motion planning with the Nao robot

B-Human 2013: Ensuring Stable Game Performance

GRASPY – Object Manipulation with NAO

Team Description for RoboCup 2012