In this paper, we propose the multi-contact locomotion planning framework for humanoid robots that leverages the target contact selection considering its long-term preferability by reinforcement learning (RL) in optimization-based motion generation with feasibility constraints. It is difficult to predict how the next target contact will affect the motion of the robot over the future in multi-contact locomotion, where humanoid robots are required to perform complex motions with kinematic constraints and static equilibrium. To solve this problem, we evaluate the preferability of the motion planned by the optimization-based motion planner to reach the target contact, which we define as contact preferability, as the reward for RL. This idea enabled us to train the policy to provide a contact with the large future preferability without explicitly designing its future promise by ourselves. We also propose the design of action space for RL based on the robot's reachability. We construct sets of feasible joint angles for each limb of the robot as successors and use them as the action space instead of directly managing contact poses. By defining the deterministic mapping from the successor to the target contact, the proposed framework can manage acyclic multi-contact motion where the number of contacts can be changed. We evaluate the proposed framework in three scenarios and prove that it can plan a preferable contact sequence for multi-contact locomotion with a high success rate and short computational time.

To reduce the computational cost of humanoid motion generation, we introduce a new approach to representing robot kinematic reachability: the differentiable reachability map. This map is a scalar-valued function defined in the task space that takes positive values only in regions reachable by the robot's end-effector. A key feature of this representation is that it is continuous and differentiable with respect to task-space coordinates, enabling its direct use as constraints in continuous optimization for humanoid motion planning. We describe a method to learn such differentiable reachability maps from a set of end-effector poses generated using a robot's kinematic model, using either a neural network or a support vector machine as the learning model. By incorporating the learned reachability map as a constraint, we formulate humanoid motion generation as a continuous optimization problem. We demonstrate that the proposed approach efficiently solves various motion planning problems, including footstep planning, multi-contact motion planning, and loco-manipulation planning for humanoid robots.

This article describes RHP Friends, a social humanoid robot developed to enable assistive robotic deployments in human-coexisting environments. As a use case application, we present its potential use in nursing by extending its capabilities to operate devices and tools according to the task and by enabling remote assistance operations. To meet a wide variety of tasks and situations in environments designed by and for humans, we develop a system that seamlessly integrates the slim and lightweight robot and several technologies: locomanipulation, multicontact motion, teleoperation, and object detection and tracking. We demonstrate the system's usage in a nursing application. The robot efficiently performs the daily task of patient transfer and a nonroutine task, represented by a request to operate a circuit breaker. This demonstration, held at the 2023 International Robot Exhibition (IREX), was conducted three times a day over three days.

This paper proposes a tactile sensor array to detect partial footholds and a wrench control method for biped humanoid robots using detected partial footholds. The proposed tactile sensor adopts a high-density FSR (Force Sensitive Resistor) array. It can read data within milliseconds. Filtering methods for its high-speed data which provide accurate estimation of contact areas are also proposed. They enable the computation of optimal contact forces. A wrench control method for biped humanoid robots, based on contact area information from detected partial footholds, enables stable force control even in partial foothold situations. The effectiveness of this advanced control system is evaluated through stabilizing and climbing stepladder experiments on partial footholds using an actual humanoid robot.

Automating in-store logistics processes in the retail industry poses significant challenges for robot manipulators. Contrary to warehouses, retails stores are subject to customer actions, which can imply non-standard tidying of products. This paper addresses the problem of detecting, discriminating, and accurately estimating the 6 degrees-of-freedom (6-DOF) pose of individual products, even in unexpected positions such as fallen or wrongly placed objects. The trained object detection model successfully discriminated similar-shaped objects of different brands/types commonly found in convenience stores. The detection is used to initialized the object position while several possible orientations are explored by a Fibonacci Multi-Start method. The estimated pose is then refined by a multi-scale projective Iterative Closest Point (ICP). The evaluation of the complete 6-DOF pose estimation module revealed its consistent ability to converge to the correct pose, avoiding local optima and achieving sub-millimetric precision. A working demonstration is presented, showcasing a robot rearranging a convenience store shelf. The overall system demonstrated the ability to detect fallen objects, estimate their poses, determine suitable grasping directions, and execute successful grasps. Importantly, the system's feasibility with minimal human intervention was demonstrated, allowing easy addition of new objects by convenience store employees or other stakeholders.

This paper describes the cybernetic avatar system developed by Team JANUS for connectivity, exploration, and skill transfer: the core domains targeted by the ANA Avatar XPRIZE competition, for which Team JANUS was a finalist. We used as an avatar a humanoid robot with a human-like appearance and shape that is capable of reproducing facial expressions and walking, and built an avatar control system that allowed the operator to control the avatar through equivalent mechanisms of motion; that is, by replicating the upper-body movement with naturalness and by stepping to command locomotion. In this way, we aimed to achieve high-fidelity telepresence and managed to be well evaluated from the point of view of the operator during the competition. We introduce our solutions to the integration challenges and present experimental results to asses our avatar system, together with current limitations and how we are planning to mitigate them in future work.

This study introduces a rapid motion-replanning technique driven by a semantic 3D voxel mapping system, essential for humanoid robots to autonomously navigate unknown territories through online environmental sensing. Addressing the challenges posed by the conventional approach based on polygon mesh or primitive extraction for mapping, we adopt semantic voxel mapping, utilizing our innovative Extended-Octomap (ExOctomap). This structure archives environmental normal vectors, outcomes of Euclidean Cluster Extraction, and principal component analysis within an Octree structure, facilitating an O(log N) efficiency in semantic accessibility from a position query x∈R3. This strategy reduces the 6D contact pose search to simple 3D grid sampling. Moreover, voxel representation enables the search of collision-free trajectories online. Through experimental validation based on simulations and real robotic experiments, we demonstrate that our framework can efficiently adapt multi-contact motions across diverse environments, achieving near real-time planning speeds that range from 13.8 ms to 115.7 ms per contact.

The purpose of this review is to clarify the problems and discuss the recent research to achieve multi-contact activities by humanoid robots from the viewpoint of contact planning, motion planning, and motion control. We also provide a brief discussion about the future perspective to improve the capability of humanoid robots in real-world applications. Multi-contact activities by humanoids are required when they work in unstructured environments such as large-scale manufacturing. To achieve them, recent works improve the computational efficiency of contact planners in the global search for a feasible contact sequence and tackle the problem of coupled kinematics and dynamics of the robot in whole-body motion planning. Combining them with the real-time controller to compensate for unexpected disturbances, humanoid robots have achieved complex multi-contact activities in the real world. Since it is computationally expensive to plan multi-contact motion with non-coplanar contact and acyclic contact transitions from scratch, we decompose the framework for multi-contact activities into contact planning, motion planning, and motion control. Then, we investigate the problems and overview the recent efforts to solve them in each component. Finally, we summarize the current achievements and discuss future directions to improve the capability of multi-contact activities by humanoids.

In this paper, we propose a trajectory planning framework for a robot that exploits a pre-computed database of end-effector trajectories as the guidance of optimization-based inverse kinematics. We constructed a reachable graph of a robot offline, which represents feasible end-effector paths with corresponding configurations. When performing the online trajectory planning, we applied A* search to the reachable graph to find a feasible path between input start and goal globally in the task space. Its cost function has the separated term dependent on the robot, which comes from the manipulability of configurations preserved in the reachable graph, and that is dependent on the environment. Then, we solve optimization-based inverse kinematics to generate an optimal joint trajectory while utilizing the end-effector trajectory and its corresponding configurations as the guidance to avoid local optimum. We evaluated our framework quantitatively by comparing it with existing methods to confirm that it achieved a high success rate and quality of results while suppressing its computational time. We also qualitatively proved its practicality by applying it to the material handling task in the real-world. This result shows that it improved the performance of the optimization-based inverse kinematics avoiding local optimum and applicability to the different environments of the pre-computed motion database.

In this letter, we propose a multi-contact locomotion planning framework for a humanoid robot to traverse complex environments utilizing bilateral contact forces with reasonable computational time. We hypothesize that a bilateral contact can be approximated by a pair of surface contacts, and expand the static CoM feasible region by this assumption, which we define as bSCFR. This assumption enables us to project centroidal statics with bilateral contact forces to a constraint on whole-body kinematics. Then, we formulate the whole-body inverse kinematics problem for all the discretized frames in the target contact transition as one optimization problem with bSCFR and CoM regularization, which can find feasible whole-body configurations while considering both kinematics and statics. After solving whole-body inverse kinematics, the contact forces are computed based on the centroidal statics, where appropriate bilateral contact forces are automatically generated according to the resulting CoM positions. We experimentally confirm that our proposed framework enabled HRP-5P to traverse steep stairs utilizing bilateral contact forces in the real world, and conclude that it expands the multi-contact locomotion capability of a humanoid robot.

In this letter, we propose an efficient and highly versatile loco-manipulation planning for humanoid robots. Loco-manipulation planning is a key technological brick enabling humanoid robots to autonomously perform object transportation by manipulating them. We formulate planning of the alternation and sequencing of footsteps and grasps as a graph search problem with a new transition model that allows for a flexible representation of loco-manipulation. Our transition model is quickly evaluated by relocating and switching the reachability maps depending on the motion of both the robot and object. We evaluate our approach by applying it to loco-manipulation use-cases, such as a bobbin rolling operation with regrasping, where the motion is automatically planned by our framework.

In order for a humanoid robot to perform loco-manipulation such as moving an object while walking, it is necessary to account for sustained or alternating external forces other than ground-feet reaction, resulting from humanoid-object contact interactions. In this letter, we propose a bipedal control strategy for humanoid loco-manipulation that can cope with such external forces. First, the basic formulas of the bipedal dynamics, i.e., linear inverted pendulum mode and divergent component of motion, are derived, taking into account the effects of external manipulation forces. Then, we propose a pattern generator to plan center of mass trajectories consistent with the reference trajectory of the manipulation forces, and a stabilizer to compensate for the error between desired and actual manipulation forces. The effectiveness of our controller is assessed both in simulation and loco-manipulation experiments with real humanoid robots.

In many robotic applications, the environment setting in which the 6-DoF pose estimation of a known, rigid object and its subsequent grasping is to be performed, remains nearly unchanging and might even be known to the robot in advance. In this paper, we refer to this problem as instance-specific pose estimation: the robot is expected to estimate the pose with a high degree of accuracy in only a limited set of familiar scenarios. Minor changes in the scene, including variations in lighting conditions and background appearance, are acceptable but drastic alterations are not anticipated. To this end, we present a method to rapidly train and deploy a pipeline for estimating the continuous 6-DoF pose of an object from a single RGB image. The key idea is to leverage known camera poses and rigid body geometry to partially automate the generation of a large labeled dataset. The dataset, along with sufficient domain randomization, is then used to supervise the training of deep neural networks for predicting semantic keypoints. Experimentally, we demonstrate the convenience and effectiveness of our proposed method to accurately estimate object pose requiring only a very small amount of manual annotation for training.

This paper proposes a new framework to generate 3D multi-contact locomotion with low computation cost. The proposed framework consists of (a) the derivation of the prospect centroidal dynamics by introducing a force distribution ratio, where it can be represented with a formulation similar to the inverted pendulum’s one, and (b) the development of a fast computation method for generating a 3D center-of-mass (CoM) trajectory. Then (c) the ZMP reference is modified so that feasible contact wrench can be generated by a force distribution using the centroidal dynamics with the approximated friction cone. The proposed method allows to generate a trajectory sequentially and to change the locomotion parameters at any time even under variable CoM height. Then, the contact timing of each end-effector can be adjusted to synchronize with the actual contact with the environment by shortening or extending the desired duration of the support phase. This can be used to improve the robustness of the locomotion. The validity of the proposed method is confirmed by several numerical results in dynamic simulator: a CoM motion while changing the contact timing, a multi-contact locomotion considering a transition between biped and quadruped walking on an horizontal floor to move below obstacles. Finally, we also show a climbing stairs using handrail which requires dynamic changes of unilateral and bilateral contacts.

In this paper, we propose a locomotion planning framework for a humanoid robot with an efficient footstep and whole-body collision avoidance planning, which enables the robot to traverse an unknown narrow space while utilizing its body structure like a human. The key idea of the proposed method is to reduce a large computational cost for the whole-body locomotion planning by executing global footstep planning first, which has a much smaller search space, and then performing a sequential whole-body posture planning while utilizing the resulting footsteps and a centroidal trajectory as a guide. In the global footstep planning phase, we modify bounding box of the robot based on the centroidal sway motion. This idea enables the planner to obtain appropriate footsteps for next whole-body motion planning. Then, we execute sequential whole-body collision avoidance motion planning by prioritized inverse kinematics based on the resulting footsteps and centroidal trajectory, which enables the robot to plan whole-body collision avoidance motion for each step within less than 100ms at worst. The major contribution of our paper is solving the problem of the increasing computational cost for whole-body motion planning and enabling a humanoid robot to execute adaptive locomotion planning on the spot in an unknown narrow space.

In this letter, we propose a graph-search based multi-contact locomotion planning method for humanoid robots, focusing on the sustainability of contacts as its key feature. We introduce the idea of sustainable contact area, which represents the area on which contacts can be maintained during contact transitions. This enables us to select feasible contact candidates along a given root path. Then, we compute all the possible combinations of these candidate contacts with every limb appearing at most once, which we call contact sets. The list of these contact sets can be regarded as a list of nodes in a graph structure representing transitions between sustainable contacts, which we name as the sustainable contact graph. We apply A* search on this graph, and evaluate the connectability of nodes by planning quasi-static motion sequences for their contact transitions. In this process, we locally modify the candidate contact to satisfy kinematics constraints and static equilibrium of the robot. The proposed method enables us to plan feasible contact transition motions without random sampling or manually designed contact transition models, and solves the problem of ignoring possible contact transitions, which is caused by the discretization in existing graph-search based planners. We evaluate our proposed method in both simulation and a real robot, and confirm that it contributes to improving the multi-contact locomotion abilities of a humanoid robot.

Nowadays, SLAM in the dynamic environment has become a popular topic. This problem is called dynamic SLAM where many solutions have been proposed to segment out the dynamic objects that bring errors to camera tracking and subsequent 3D reconstruction. However, state-of-the-art dynamic SLAM methods face the problems of accuracy and speed, which is due to the fact that one segmentation algorithm cannot guarantee both points at the same time. In this paper, we propose a multi-purpose dynamic SLAM framework to provide a variety of selections for segmentation, each has its applicable scene. The experimental results show that the framework is compatible with different segmentation methods, which improves corresponding existing methods in some aspects.

This paper discusses a lateral balance controller for a biped robot with both legs fully extended. In a conventional position-controlled legged robot, a balance control with stretched knees is an open problem since the mechanical singularity prevents the direct control of the floor force distribution. To control forces indirectly, we introduce an additional acceleration of the center of mass and a ZMP modification as control inputs. The lateral balance controller is designed as a state feedback system by using a data driven approach. The proposed lateral controller was merged with a sagittal controller based on the Spatially Quantized Dynamics (SQD), then it helped our humanoid robot HRP-2Kai to achieve laterally well balanced, knee-stretched, and long stride gait.

Twenty-three years have passed since Honda unveiled its P2 robot in 1996. Even though extensive research has been conducted since then, humanoid robots have not been commercialized for practical applications. It is important to carefully select target applications to commercialize humanoid robots as soon as possible and solve social issues, such as labor shortages in industries and elderly care. This article consists of two parts. First, we discuss what we consider to be the most promising application of the commercialization of humanoid robots, based on our experience in their development. Several candidate applications are evaluated on four criteria, and we conclude that large-scale manufacturing is the most promising one.

This letter introduces the humanoid robot HRP-5P, which stands for Humanoid Robotics Platform-5 Prototype. We have been developing the HRP series humanoid robots since 2000, and HRP-5P is the latest version of the HRP series as of 2018. It is developed as a prototype for our next generation humanoid robotics platform, aiming to realize the use of practical humanoid robots in place of humans within large-scale assembly industries such as construction sites, aircraft facilities, and shipyards. To realize it, electrically actuated high-power joints with wide movable range have been newly designed. Also, the arm configuration has also been redesigned to improve the physical ability of working on actual sites. The mechanism and the electrical systems are presented with its basic specification in this letter.