Felix Sygulla

Fast object approximation for real-time 3D obstacle avoidance with biped robots

In order to achieve fully autonomous humanoid navigation, environment perception must be both fast enough for real-time planning in dynamic environments and robust against previously unknown scenarios. We present an open source, flexible and efficient vision system that represents dynamic environments using simple geometries. Based only on onboard sensing and 3D point cloud processing, it approximates objects using swept-sphere-volumes while the robot is moving. It does not rely on color or any previous models or information. We demonstrate the viability of our approach by testing it on our human-sized biped robot Lola, which is able to avoid moving obstacles in real-time while walking at a set speed of 0.4m/s and performing whole-body collision avoidance.

Real-time predictive kinematic evaluation and optimization for biped robots

Collision-free walking in cluttered environments is still an open issue for humanoids. Most current approaches use heuristics with large safety margins to plan the robots motion. That way, the chance of collisions can be greatly reduced but the robot movements are artificially limited. In this context, we extend our framework for motion generation and whole-body collision-avoidance by an on-line predictive kinematic parameter evaluation and optimization: we propose to evaluate the parameter set describing the walking pattern by integrating the full kinematic model of the robot. Initial parameter sets, which are kinematically infeasible due to kinematic limits or collisions, can be identified and adapted before the motion is executed. Starting with a feasible solution, the parameter set is optimized using a gradient method. Since the method is applied before each step, while the robot is executing the previous step, it is very reactive to changes in the environment or in the user input. The optimization method is not limited to a specific walking pattern representation, but it is applicable to different representations. We want to emphasize its suitability for real-time control. The optimization can be stopped if it exceeds a predetermined time budget. In that case, an executable but suboptimal result is used. The method is presented with simulation results obtained with our multi-body simulation. We have also validated the real-time performance in experiments with our humanoid Lola.

Model-based predictive bipedal walking stabilization

A well known strategy in bipedal locomotion to prevent falling in the presence of large disturbances is to modify drastically future motion. This is an important capability of a walking control system in order to bring humanoid robots from controlled laboratory conditions to real environment situations. This paper presents a predictive stabilization method which modifies planned center of mass and foot trajectories depending on the current state of the robot. It uses a nonlinear prediction model [1] and applies a conjugate gradient method to solve the resulting optimization problem in real-time. Furthermore, the method is integrated in the walking control system of our bipedal robot LOLA. Simulation results demonstrate the effectiveness and the advantages of the proposed method.

Motion planning for redundant manipulators in uncertain environments based on tactile feedback

The exploitation of new fields of application in addition to traditional industrial production for robot manipulators (e.g. agriculture, human areas) requires extensions to the sensor as well as to the planning capabilities. Motion planning solely based on visual information performs poorly in cluttered environments since contacts with obstacles might be inevitable and thus a distinction between hard and soft objects has to be made. In our contribution we present a novel intrinsic tactile sensing module mounted on a multipurpose 9 DOF agricultural manipulator. With its innovative sensor arrangement we consider it to be a low-cost, easily manageable and efficient solution with a reasonable abstraction layer in comparison to complex torque sensing or tactile skins. The sensor provides information about the resulting force and torque. In the second part of our paper, the tactile information is used for minimizing contact forces while pursuing the end-effector tasks as long as reasonable. Hence, we present robust and efficient extensions to Resolved Motion Rate Control for real-time application. We introduce a general formulation providing control inputs in task-space, joint-space and nullspace. Thus, we design a suitable controller by feedback linearization and feed-forward terms. Results from real-world experiments show the potential of our approach. A discussion of the different control schemes completes the paper.

Time-variable, event-based walking control for biped robots

Most walking controllers for biped robots are based on a synchronized phase-based structure, where trajectories are executed following predefined timing constraints. This inherent fixed time dependency makes humanoid robots extremely susceptible to irregularities in terrain compared to their biological counterparts. We present an event-based control strategy which incorporates a time-variable phase to help deal with unexpected early and late contact situations. It results in an improved robustness against such scenarios, as shown by simulation results of our robot Lola.

Real-Time Path Planning in Unknown Environments for Bipedal Robots

Autonomous navigation in dynamic and unknown environments requires real-time path planning. Solving the path planning problem for bipedal locomotion quickly and robustly is one of the main challenges in making humanoid robots competitive against mobile platforms. In this letter, we propose strategies to use mobile platform planners for improving the navigation of bipedal robots. These strategies combine advantageously continuous two-dimensional (2-D) paths with conventional step planners for humanoid robots. We introduce a mobile platform planner suitable for real-time navigation. It searches for multiple 2-D paths that makes the path planning more robust against limited calculation time and changing scenarios. It is combined with a step planner and integrated in the framework for autonomous navigation of our robot Lola. We evaluate different strategies in simulation and validate them in experiments in unknown dynamic environments.

A flexible and low-cost tactile sensor for robotic applications

For humans, the sense of touch is essential for interactions with the environment. With robots slowly starting to emerge as a human-centric technology, tactile information becomes increasingly important. Tactile sensors enable robots to gain information about contacts with the environment, which is required for safe interaction with humans or tactile exploration. Many sensor designs for the application on robots have been presented in literature so far. However, most of them are complex in their design and require high-tech tools for their manufacturing. In this paper, we present a novel design for a tactile sensor that can be built with low-cost, widely available materials, and low effort. The sensor is flexible, may be cut to arbitrary shapes and may have a customized spatial resolution. Both pressure distribution and absolute pressure on the sensor are detected. An experimental evaluation of our design shows low detection thresholds as well as high sensor accuracy. We seek to accelerate research on tactile feedback methods with this easy to replicate design. We consider our design a starting point for the integration of multiple sensor units to a large-scale tactile skin for robots.

Modifying the estimated ground height to mitigate error effects on bipedal robot walking

Classic biped walking controllers assume a perfectly flat, rigid surface on which the robot walks. While walking over unknown terrain, robots need to sense and estimate the ground location. Errors in this estimation result in an unexpected early or late ground contact of the swing foot. In this paper, we analyze how these errors affect walking stability. Based on simulation results, we propose a strategy that mitigates this effect. We show that if the ground height has an associated uncertainty, an overestimation of its value results in a more stable walk. This overestimation depends on both sensor data and the robots dynamics. By using a reduced robot model, our strategy could be implemented into the real-time control to make the robot more robust against perception errors and irregular surfaces.

Adaptive motion control in uncertain environments using tactile feedback

Next generation robot applications are expected to leave the field of complex tasks in simple environments and move on to simple and complex tasks in complex environments. In our opinion, tactile feedback is a key technology for motion planning in such unstructured environments as visual information may be insufficient or even unavailable. In this paper, we show the performance of a tactile feedback controller in joint-space, which is not bound to the null space of the manipulator. Additionally, we extend our tactile feedback control framework to hierarchical multi-space controllers with adaptive prioritization. This allows to dissolve the trade-off between low contact forces and good positional tracking and aims at applications, where desired trajectories must be held using manipulator redundancy and end-effector deviation is only admissible at high contact forces. The stability of this approach is discussed as well. Furthermore, we present an online stiffness estimation algorithm to increase the performance of our controllers in uncertain environments. Several real-world experiments with a 9-DOF multipurpose manipulator in collision with soft and hard objects show the capability of our work.