Youngbum Jun

Continuous trajectory optimization for autonomous humanoid door opening

The upcoming DARPA Robotics Challenge (DRC) presents a demanding set of real-world tasks to be accomplished autonomously by robots. In this paper, we describe the development of a system to control an existing humanoid robot to open a door, one of the many tasks of the DRC. Special emphasis is placed upon generating smooth trajectories which minimize unnecessary motion of the robot. We describe methods for generating and optimizing trajectories for the robot, and present preliminary results demonstrated on the physical robotic platform. To the best of our knowledge, we demonstrate the first large scale application of the CHOMP trajectory optimization in a situation with closed kinematic chain constraints.

DRC-hubo walking on rough terrains

Up to now humanoid robots have been designed primarily for walking on flat surfaces. In the future, humanoid robots are required to replace human beings to operate in natural or damaged man-engineered environments. In the 2013 DARPA Robotics Challenge, the robots are required to walk through several type of rough terrains. In this scenario, the robot will be challenged to keep balance and fulfill the tasks while walking. We have developed several balance gaits and associated controllers. The latter collaborate with a computer vision system to enable our humanoid robot DRC-Hubo to walk over rough terrains. Both theoretical and experimental results are presented to verify the approach.

Planning complex physical tasks for disaster response with a humanoid robot

Humanoid robots are attempting ever more complex tasks in lieu of humans. Disaster response is a promising area for the use of humanoids due to safety concerns. However, controlling a high DOF humanoid robot to autonomously perform a complex task in unknown and unstructured environments is challenging. In this paper we describe a simulation framework for humanoid grasping and transport tasks that includes dynamics, and is easily ported to a real physical humanoid robot. The system can be used to rapidly prototype humanoid motions and dynamics in simulation, and can then be ported to the physical hardware. Experimental results are presented, both in simulation and physical experiments, with the HUBO humanoid, on a task from the DARPA Robotics Challenge, attaching a fire hose to a hydrant.

Realization of miniature humanoid for obstacle avoidance with real-time ZMP preview control used for full-sized humanoid

Many walking pattern generators for humanoid robots require predefined trajectories for the robot to track. This inflexibility limits the range of real-world environments that the robot can navigate through. For environments with obstacles and inconsistent terrain, the ability to change the walking trajectory becomes valuable. Using a miniature humanoid, a three-dimensional inverted pendulum model and ZMP preview control with ZMP and Foot generator were used to implement a real-time ZMP preview controller. We show the simulation results walking on obstacle field which validates that this approach can generate the all types of walking pattern based on the distance estimated based on the sensor data to the target to step without the predefined trajectory.

A 3-Tier Infrastructure: Virtual-, Mini-, Online-Hubo Stair Climbing as a Case Study

This paper introduces a 3-tier infrastructure for humanoid research. Using the KAIST humanoid Hubo-2, virtual, mini-, and online-Hubo comprise an infrastructure to respectively prototype, test-and-evaluate, and verify-andvalidate algorithms. The resulting closed-loop design cycle promotes research that can reproduced and verified by the humanoid research community. This paper presents stair-climbing as a case study, thus demonstrating the efficacy of the 3-tier approach. Beyond the 3-tier infrastructure, the paper presents an effective climbing pattern generation and analytic inverse kinematics. This stairclimbing approach is prototyped in virtual-Hubo, experimentally tested-and-evaluated on mini-Hubo, and verifiedand-validated on online-Hubo.

Real-time teleop with non-prehensile manipulation

In this work, we present a framework for teleoperation of manipulation tasks under low bandwidth, high latency conditions. This framework allows us to combine multiple manipulation and walking strategies to quickly adapt to changing mission parameters and conditions. In particular, this framework addresses the challenges of the hose attachment task of the DARPA Robotics Challenge, which encompasses walking with drag, grasping in constrained environments, and complex, close chain manipulation.

Strategies for driving and egress for the vehicle of a humanoid robot in the drc finals 2015

This paper presents various strategies for humanoid vehicle driving and egress tasks. For driving, a tele-operating system that controls a robot based on a human operator’s commands is built. In addition, an autonomous assistant module is developed for the operator. Normal position control can result in severe damage to robots when they egress from vehicles. To prevent this problem, another approach that mixes various joint control techniques is adopted in this study. Additionally, a footplate is newly designed and attached to the vehicle floor for the ground landing phase of the egress task. The attached plate enables the robot to step down onto the ground in a safe manner. For stable locomotion, a balance controller is designed for the humanoid. For the design of the controller, the robot is modeled using an inverted pendulum that consists of a spring and a damper. Then, a state feedback controller (with pole placement and a state observer) is built based on the simplified model. Many approaches that are presented in this paper were successfully applied to a full-sized humanoid, DRC-HUBO+, in the DARPA Robotics Challenge Finals, which were held in the United States in 2015.

A hardware-in-the-loop test rig for aerial manipulation

A hardware-in-the-loop test rig is presented to bridge the gap between basic aerial manipulation research and the ability of flying robots to perform tasks such as door opening, bridge repair, agriculture care, and other applications requiring interaction with the environment. Unmanned aerial vehicles have speed and mobility advantages over ground vehicles and can operate in 3-dimensional workspaces. In particular, the usefulness of these capabilities is highlighted in areas where ground robots cannot reach or terrains they are unable to navigate. However, most UAVs operating in nearearth or indoor environments still do not have the payload capabilities to support multi-degree of freedom manipulators. We present a rotorcraft emulation environment using a 7 degree of freedom manipulator. Since UAVs require significant setup time and to avoid potential crashes, our test and evaluation environment provides repeatable experiments and captures reactionary forces experienced during ground interaction. Our preliminary results indicate that we can accurately model, emulate, and control our aircraft-manipulator system during both arm actuation and interacting with target objects.

Controlling and maximizing humanoid robot pushing force through posture

Pushing is one of many object manipulation strategies that requires interaction with the environment. Many force control approaches have been proposed for such manipulation. In a force controller implementation for a humanoid robot, however, there is no fixed base. If the required reaction force is greater than the humanoid robot can support, the robot will lose its balance. This paper presents a method to expand these force limits by changing a humanoid robots posture. Based on Double Inverted Pendulum (DIP) model, the force limitation that the humanoid robot can support is calculated. With a feet-apart strategy and whole-body posture, a method is proposed to maximize the force limitation under the condition that the height of the target object is constant. Finally, comparison of simulation and experimental data validates the approach.

Team DRC-Hubo@UNLV in 2015 DARPA Robotics Challenge Finals

This chapter presents a technical overview of Team DRC-Hubo@UNLVs approach to the 2015 DARPA Robotics Challenge Finals (DRC-Finals). The Finals required a robotic platform that was robust and reliable in both hardware and software to complete tasks in 60 min under degraded communication. With this point of view, Team DRC-Hubo@UNLV integrated methods and algorithms previously verified, validated, and widely used in the robotics community. For the communication aspect, a common shared memory approach that the team adopted to enable efficient data communication under the DARPA controlled network is described. A new perception head design (optimized for the tasks of the Finals) and its data processing are then presented. In the motion planning and control aspect, various techniques, such as wheel-driven navigation, zero-moment point (ZMP)-based locomotion, and position-based manipulation and controls, are described in this chapter. By introducing strategically critical elements and key lessons learned from DRC-Trials 2013 and the testbed of Charleston, we also illustrate how DRC-Hubo has evolved successfully toward the DRC-Finals.