One of the greatest challenges in robot control is closing the gap between themotion capabilities of humans and humanoid robots. The difficulty lies in thecomplexity of the dynamical systems representing the said robots: theirnonlinearity, underactuation, discrete behavior due to collisions and friction,high number of degrees of freedom. Moreover, humanoid robots are supposed tooperate in non-deterministic environments, which require advanced real timecontrol. The currently prevailing approach to coping with these difficulties isto impose various limitations on the motions and employ approximate models ofthe robots. In this thesis, we follow the same line of research and propose anew approach to the design of balance preserving whole body motion controllers.The key idea is to leverage the advantages of whole body and approximate modelsby mixing them within a single predictive control problem with strictlyprioritized objectives.Balance preservation is one of the primary concerns in the control of humanoidrobots. Previous research has already established that anticipation of motionsis crucial for this purpose. We advocate that anticipation is helpful in thissense as a way to maintain capturability of the motion, i.e., the ability tostop. We stress that capturability of anticipated motions can be enforced withappropriate constraints. In practice, it is common to anticipate motions usingapproximate models in order to reduce computational effort, hence, a separatewhole body motion controller is needed for tracking. Instead, we propose tointroduce anticipation with an approximate model into the whole body motioncontroller. As a result, the generated whole body motions respect thecapturability constraints and the anticipated motions of an approximate modeltake into account whole body constraints and tasks. We pose our whole bodymotion controllers as optimization problems with strictly prioritizedobjectives. Though such prioritization is common in the literature, we believethat it is often not properly exploited. We, therefore, propose severalexamples of controllers, where prioritization is useful and necessary toachieve desired behaviors. We evaluate our controllers in two simulatedscenarios, where a whole body task influences walking motions of the robot andthe robot optionally exploits a hand contact to maintain balance whilestanding.

Research on humanoid robots has been mostly with the aim of developing robots that can replace humans in the performance of certain tasks. Motion planning for these robots can be quite difficult, due to their complex kinematics, dynamics and environment. It is consequently one of the key research topics in humanoid robotics research and the last few years have witnessed considerable progress in the field. Motion Planning for Humanoid Robots surveys the remarkable recent advancement in both the theoretical and the practical aspects of humanoid motion planning. Various motion planning frameworks are presented in Motion Planning for Humanoid Robots, including one for skill coordination and learning, and one for manipulating and grasping tasks. The problem of planning sequences of contacts that support acyclic motion in a highly constrained environment is addressed and a motion planner that enables a humanoid robot to push an object to a desired location on a cluttered table is described. The main areas of interest include: • whole body motion planning, • task planning, • biped gait planning, and • sensor feedback for motion planning. Torque-level control of multi-contact behavior, autonomous manipulation of moving obstacles, and movement control and planning architecture are also covered. Motion Planning for Humanoid Robots will help readers to understand the current research on humanoid motion planning. It is written for industrial engineers, advanced undergraduate and postgraduate students.

This book aims at providing algorithms for balance control of legged, torque-controlled humanoid robots. A humanoid robot normally uses the feet for locomotion. This paradigm is extended by addressing the challenge of multi-contact balancing, which allows a humanoid robot to exploit an arbitrary number of contacts for support. Using multiple contacts increases the size of the support polygon, which in turn leads to an increased robustness of the stance and to an increased kinematic workspace of the robot. Both are important features for facilitating a transition of humanoid robots from research laboratories to real-world applications, where they are confronted with multiple challenging scenarios, such as climbing stairs and ladders, traversing debris, handling heavy loads, or working in confined spaces. The distribution of forces and torques among the multiple contacts is a challenging aspect of the problem, which arises from the closed kinematic chain given by the robot and its environment.

Robustly maintaining balance on two legs is an important challenge for humanoid robots. The work presented in this book represents a contribution to this area. It investigates efficient methods for the decision-making from internal sensors about whether and where to step, several improvements to efficient whole-body postural balancing methods, and proposes and evaluates a novel method for efficient recovery step generation, leveraging human examples and simulation-based reinforcement learning.

This thesis aims at providing a solution to the problem of motion generation for humanoid robots. The proposed framework generates whole-body motion using the complete robot dynamics in the task space satisfying contact constraints. This approach is known as operational-space inverse-dynamics control. The specification of the movements is done through objectives in the task space, and the high redundancy of the system is handled with a prioritized stack of tasks where lower priority tasks are only achieved if they do not interfere with higher priority ones. To this end, a hierarchical quadratic program is used, with the advantage of being able to specify tasks as equalities or inequalities at any level of the hierarchy. Motions where the robot sits down in an armchair and climbs a ladder show the capability to handle multiple non-coplanar contacts. The generic motion generation framework is then applied to some case studies using HRP-2 and Romeo. Complex and human-like movements are achieved using human motion imitation where the acquired motion passes through a kinematic and then dynamic retargeting processes. To deal with the instantaneous nature of inverse dynamics, a walking pattern generator is used as an input for the stack of tasks which makes a local correction of the feet position based on the contact points allowing to walk on non-planar surfaces. Visual feedback is also introduced to aid in the walking process. Alternatively, for a fast balance recovery, the capture point is introduced in the framework as a task and it is controlled within a desired region of space. Also, motion generation is presented for CHIMP which is a robot that needs a particular treatment.

A major obstacle that prevents humanoid robots from accomplishing real world tasks is their inability to physically interact with, and effectively manipulate, the most common objects generally found in human environments. Even tasks that seem simple for a human remain a significant challenge for most robots. Robots generally employ precision to perform a manipulation task. Humans, in contrast, employ compliance through tactile and force feedback to overcome their imprecision, allowing them to resolve uncertainties associated with the task. The lack of compliance and force control has been indeed a major limiting factor in the ability of robots to interact and manipulate in human environments. One of the major objectives of this research is to endow humanoid robots with whole-body compliant motion abilities. With compliance, a robot overcomes position uncertainties by moving in directions that reduce contact forces, which in turn directs it towards its goal. Whole-body framework was designed to allow the robot to compliantly interact with its environment at multiple contact points. The synthesis of compliant tasks is greatly simplified by being independent of postures and constraints, which are automatically integrated in the control hierarchy. This research focuses on the development of (I) sensor-based whole-body compliant motion primitives, (II) contact sensing and contact force control, (III) whole-body multi-contact for extended support, kneeling, crawling, leaning table, and locomotion strategy to improve support in unstructured terrains, (IV) dynamic collision-free motion planning and (V) dynamic collision-free walking path planning.

Academic Paper from the year 2017 in the subject Engineering - Robotics, language: English, abstract: Researchers dream of developing autonomous humanoid robots which behave/walk like a human being. Biped robots, although complex, have the greatest potential for use in human-centered environments such as the home or office. Studying biped robots is also important for understanding human locomotion and improving control strategies for prosthetic and orthotic limbs. Control systems of humans walking in cluttered environments are complex, however, and may involve multiple local controllers and commands from the cerebellum. Although biped robots have been of interest over the last four decades, no unified stability/balance criterion adopted for stabilization of miscellaneous walking/running modes of biped robots has so far been available. The literature is scattered and it is difficult to construct a unified background for the balance strategies of biped motion. The zero-moment point (ZMP) criterion, however, is a conservative indicator of stabilized motion for a class of biped robots. Therefore, we offer a systematic presentation of multi-level balance controllers for stabilization and balance recovery of ZMP-based humanoid robots.