Towards general purpose service robots: World Robot Summit – Partner Robot ChallengeContreras, Luis; Yamamoto, Takashi; Matsusaka, Yosuke; Okada, Hiroyuki
doi: 10.1080/01691864.2022.2109428pmid: N/A
Although the skills required to solve specific robotics problems are reaching amazing performances recently, we propose the evaluation of such individual solutions in fully integrated robot systems tested in real daily situations like those presented at international robotics competitions. The World Robot Summit (WRS)– Partner Robot Challenge surges from the necessity to standardise and spread the research on Domestic Service Robots where a series of solutions can be tested to solve a general purpose task in a standard domestic environment. We propose the Tidy-Up task as a benchmark and present the key features behind this challenge as well as our main findings through the several stages in the design process. This approach has been proven successful in different international competitions, namely, the RoboCup Japan Open, the Mexican Tournament of Robotics, the RoboCup Worldwide and the WRS.
Sparse-Map: automatic topological map creation via unsupervised learning techniquesHernández, Jesús; Savage, Jesús; Negrete, Marco; Contreras, Luis; Sarmiento, Carlos; Fuentes, Oscar; Okada, Hiroyuki
doi: 10.1080/01691864.2022.2114296pmid: N/A
Most robots use 2D occupancy grid maps for navigation, localization, and path-planning. This model is flexible and allows to represent any geometrical shape with finite accuracy. However, this dense representation imposes high memory requirements and does not generalize well to 3D environments. We present a task-based map compression technique useful for path-planning and navigation in indoor environments for service robots where, from a point cloud of 3D map features, we calculate a number of clusters based on their spatial position and generate a sparse 3D representation of the environment. Moreover, we propose several metrics to assess the quality and performance of a map representation, and we tested our proposal using a series of point-cloud benchmarks and clustering techniques where our method has a comparable performance using a fraction of the memory footprint than the baselines. Finally, we have released our system as a Robot Operating System (ROS) based open source library.
Solution of World Robot Challenge 2020 Partner Robot Challenge (Real Space)Ono, Tomohiro; Kanaoka, Daiju; Shiba, Tomoya; Tokuno, Shoshi; Yano, Yuga; Mizutani, Akinobu; Matsumoto, Ikuya; Amano, Hayato; Tamukoh, Hakaru
doi: 10.1080/01691864.2022.2115315pmid: N/A
The World Robot Challenge is an international competition for the social implementation of robots. Among them, the Partner Robot Challenge (Real Space) is a category that focuses on domestic service robots, competing to achieve simple tasks such as tidying a room, avoiding small obstacles, grasping a specified object from the shelf, and delivering an object to a waving person. In this category, we focused on the theme of ‘Keep Moving’ and worked on researching and developing technologies for object recognition, grasping, and other tasks. For object recognition, we propose an automatic dataset generator using a physics simulator and generate about 490,000 images in two hours. We trained it with an instance segmentation model, achieving an accuracy of about 79.3% in all games. For object grasping, we installed a three-dimensional sensor on the robot hand, and achieved successful grasping with high accuracy of about 79.6% (88.9% excluding failures due to hardware problems) in all games. In addition, we propose a motion synthesis method that simultaneously performs movement and posture transition to achieve high-speed motion. When navigation is executed, the robot changes to the navigation posture while aiming for the goal. Before reaching the goal, it changes the posture for the following action. As a result, we realized a speedup of about 1.32 times in the pick-and-place task. In the tidy-up task, the robot could clear an average of 14 objects and a maximum of 16 objects in 15 minutes. In the obstacle avoidance task, the robot succeeded five times out of six games, and in the delivery task, it succeeded four times out of six games. Throughout all the tasks, we scored 800 points twice, the highest score in the world, and won the championship. This research is aimed at the practical application of domestic service robots. We believe that the results presented in this paper demonstrate the practicality of domestic service robots in a well-developed environment. The video of the finals is available at https://youtu.be/ElUb8bfSC34?t=5511 (right side). The our source code using the simulator is available at https://github.com/Hibikino-Musashi-Home/hma_wrs_sim_ws.
Selective instance segmentation for pose estimationMatsumoto, Kazuhisa; Ibuki, Yusuke; Tomikawa, Ryusei; Kobayashi, Kazufumi; Ohara, Kenichi; Tasaki, Tsuyoshi
doi: 10.1080/01691864.2022.2104621pmid: N/A
Object detection and pose estimation are required for automating the stocking of shelves in retail stores and improving pose estimation accuracy necessitate instance segmentation. However, conventional methods experience difficulties in moving the robot in real time because they have too many parameters, which increases the processing time. In this study, we developed a high-speed instance segmentation method that solves this problem. Specifically, we focused on the fact that the robot's task target is a single object. Consequently, by choosing one detection target in the robot's view (image), we reduce the size required by the deep neural network and accelerate instance segmentation. We used the attention region as our selection method because it does not increase the number of parameters. Further, by using an instance in the attention region, we selectively output only high-precision instances. The results of experiments conducted showed a 12.1 pt improvement in instance segmentation precision and 2.5 times faster execution on a CPU compared with previous methods, as well as a 38.4 pt improvement in pose estimation precision.
Verification experiment of stocking and disposal tasks by automatic shelf and mobile single-arm manipulation robotTanaka, Junya; Yamamoto, Daisuke; Nakashima, Ryo; Yamaji, Yuto; Ohtsu, Hiroshi; Kamata, Keisuke; Nara, Kohei; Asayama, Tetsuya
doi: 10.1080/01691864.2022.2104622pmid: N/A
This study proposes a stock handling system that combines an automatic shelf and a picking robot with the aim of reducing the labor required for stocking and disposing of items. The robot system was developed by referring to the rules of the robot competition WRS2020 F.C.S.C. The picking robot is a standard single-arm mobile manipulation robot and the automatic shelf is characterized by the fact that the shelves move up and down and back and forth automatically. Automatic movement of the shelves in the pull-out direction allows the end-effector of the manipulator to approach the item from above and select the grasping position from a number of candidates. This improves the stability of the grasping of the item and is expected to prevent the item from falling during transportation. When there are tall items at the front of the shelf, the end-effector can directly approach the items at the back of the shelf from above the shelf as the shelf plate automatically moves in the pull-out direction. We report the mechanical configuration of the automatic shelf and basic experimental results for stocking and disposal work by the proposed robot system.
A robot hand with free-rotating grips for switching three different grasping modesSeki, Masashi; Wada, Kazuyoshi; Teraguchi, Tomoya; Tomizawa, Tetsuo
doi: 10.1080/01691864.2022.2114298pmid: N/A
The automation of the display and disposal of products on shelves has attracted widespread attention because it is the most time-consuming task in convenience store operations. However, robots find it difficult to perform store operations because products vary in shapes and sizes and must be displayed in specific positions on narrow shelves. In this study, we developed a robot hand that can grasp various products and easily change its posture. The hand has two free rotating grips that allows the robot to change position between the following three grasping modes: (i) grasp mode: prevents the free rotation of an object, (ii) rotation mode: changes the posture of the object due to its weight, and (iii) scoop mode: allows the robot to grasp a flat object. A performance evaluation experiment was conducted using the robot hand. Subsequently, three results were obtained. First, the robot hand successfully grabbed a cork block, a rice ball, and a sandwich in the grasp and rotation mode, and the gripper behavior changed depending on the grasping mode. Second, the robot hand successfully grasped a lunch box in the scoop mode. Third, the robot hand switched from the grasp to the rotation mode while grasping the object. These results indicate that the hand can change its posture owing to gravity at any time.
Robot behavior debugger for non-expert users in convenience stores using behavior treesTulathum, Pattaraporn; Usawalertkamol, Bunyapon; Garcia Ricardez, Gustavo Alfonso; Takamatsu, Jun; Ogasawara, Tsukasa; Matsumoto, Kenichi
doi: 10.1080/01691864.2022.2115316pmid: N/A
As the problematic consequence of an aging society, with increasing labor shortage, there is a need for service robots to efficiently support the works in many places such as convenience stores. However, it is difficult to program a robot behavior and meet the needs of the shop staff (i.e. non-expert users who lack of knowledge and experience in robot programming). Hence, there is the need for a system to help non-expert users to identify and fix the issues within the robot behaviors. This paper proposes a Behavior Tree-based robot behavior creation system for non-expert users with four debugging features and a simulator. Non-expert users can use drag-and-drop composition to create the robot behavior program. Moreover, our debugger allows non-expert users to use breakpoints, log node status, monitor node execution, show robot status variables, and also can visually verify the robot behaviors via a simulator. We evaluate the effectiveness of our debugging system with 14 non-expert users by asking them to solve three tasks (i.e. creating and fixing Behavior Trees) from the given convenience store scenario. The experimental results show that more than 70% of non-expert users can utilize our debugging features to finish all the tasks. Additionally, our system usability has a high marginal level from the subjects' perspective according to the System Usability Scale (SUS).