Taskin Padir

Team WPI-CMU: Achieving Reliable Humanoid Behavior in the DARPA Robotics Challenge

In the DARPA Robotics Challenge DRC, participating human-robot teams were required to integrate mobility, manipulation, perception, and operator interfaces to complete a simulated disaster mission. We describe our approach using the humanoid robot Atlas Unplugged developed by Boston Dynamics. We focus on our approach, results, and lessons learned from the DRC Finals to demonstrate our strategy, including extensive operator practice, explicit monitoring for robot errors, adding additional sensing, and enabling the operator to control and monitor the robot at varying degrees of abstraction. Our safety-first strategy worked: we avoided falling, and remote operators could safely recover from difficult situations. We were the only team in the DRC Finals that attempted all tasks, scored points 14/16, did not require physical human intervention a reset, and did not fall in the two missions during the two days of tests. We also had the most consistent pair of runs.

Achieving reliable humanoid robot operations in the DARPA robotics challenge: Team WPI-CMU’s approach

The DARPA Robotics Challenge (DRC) required participating human-robot teams to integrate mobility, manipulation, perception and operator interfaces to complete a simulated disaster mission. We describe our approach to the development of manipulation and locomotion capabilities for the humanoid robot atlas unplugged developed by Boston Dynamics. We focus on our approach, results and lessons learned from the DRC Finals to demonstrate our strategy including extensive operator practice, explicit monitoring for robot errors, adding additional sensing, and enabling the operator to control and monitor the robot at varying degrees of abstraction. Our safety-first strategy worked: we avoided falling and remote operators could safely recover from difficult situations. We were the only team in the DRC Finals that attempted all tasks, scored points (14/16), did not require physical human intervention (a reset), and did not fall in the two missions during the two days of tests. We also had the most consistent pair of runs. We ranked 3rd out of 23 teams when the scores from two official runs were averaged.

Towards Autonomous Grasping with Robotic Prosthetic Hands

For an effective use of a robotic prosthetic, smooth transition between the states of this human-in-the-loop cyber-physical system (or autonomous operation) is crucial. In order to achieve that, taking into account the uncertainties in the robotic system may be as significant as anticipating the human intent. In this study, we present two risk measures associated with a simple grasping task based on the pre-grasp position and orientation of the robotic hand, and analyze them by using a model of a dexterous mechatronic prosthetic arm to simulate a simple grasping task. Moreover, we develop a hybrid position and force grasp controller and use a grasp quality metric to assess the grasps. 34 simulations with different pre-grasp poses were performed to investigate the relationship between the grasp quality and the risk measures as well as the sensitivity of the grasp quality to the pre-grasp position of the hand. Results show that the proposed risk metrics and the pre-grasp position correlate with the grasp quality, yet they are not sufficient to predict the outcome.

CWave: High-performance single-source any-angle path planning on a grid

Path planning on a 2D-grid is a well-studied problem in robotics. It usually involves searching for a shortest path between two vertices on a grid. Single-source path planning is a modified problem which asks to find distances from a given point to all other points on the map. A high-performance algorithm for single-source any-angle path planning on a grid that we named CWave is proposed in this work. “Any-angle” attribute of a path planning algorithm implies that such algorithm can find paths which may include any angle segments, as opposed to standard A∗ on an 8-connected graph, the path can turn with 45°-increments only. The key idea of the presented algorithm is that it does not represent the grid as a graph and uses discrete geometric primitives to define the wave front. In its purest form, CWave requires for computation only integer arithmetics and multiplication by two, but can accumulate the distance error at turning points. A modified version of CWave with minimal usage of floating-point calculations is also developed. It allows to eliminate any accumulative errors which is proven mathematically and experimentally on several maps. The performance of the algorithm on three maps is demonstrated to be significantly faster than that of Theta∗, Lazy Theta∗ and Field A∗ adapted for single-source planning. The limitations of the current implementations of the algorithm as well as potential improvements are discussed.

LocATER: Localization and accountability technologies for emergency responders

Over the last decade, there have been several unsuccessful attempts to commercialize indoor localization technology for emergency responders mainly due to a poor product-market fit. This paper describes in detail the findings and conclusions of NSF Innovation Corps (I-Corps) Team 735 (Customer Discovery for Field-Deployable Indoor Localization Technology), part of the Spring 2016 DC-area cohort. As part of the I-Corps curriculum, the team interviewed over 100 relevant stakeholders to better understand the challenges facing the successful commercialization of indoor localization technologies. Throughout the course of these interviews, it became evident that the current solutions available commercially and in research environments did not meet all the requirements needed for effective response in challenging indoor environments. In addition, it is clear that previously demonstrated localization capabilities are sufficient to meet the technical requirements demanded by emergency response scenarios, but have previously not been unified in a holistic system that can be successfully introduced to the market.

Template-based human supervised robot task programming

Motions of a robot interacting with its environment can be described by a set of constraints. This paper introduces an approach, called motion template, which can quickly program and compose the constraints for the motion planner to generate the trajectory. Two types of motion templates, grasp and turn, are specifically described to explain the details of the technique. The reusability and shareability properties of the motion template are demonstrated using a variety of the motion planning applications across different robot platforms. A motion template framework is used to implement the motion template with the trajectory optimization.

Task-oriented planning algorithm for humanoid robots based on a foot repositionable inverse kinematics engine

Two major questions humanoid robots need to solve for manipulation tasks are whether they need to take steps and where to take steps to. In this paper, we introduce the formulation of a powerful inverse kinematics (IK) engine which can help to answer these questions. In the engine, the IK problem is formulated as an optimization problem. After configuring appropriate costs and constraints, the IK engine can compute a solution in which the robot is allowed to reposition its feet if necessary for meeting the task requirements. When the answers of these two questions are found, an integrated planning algorithm involving an IK solver, footstep planning and whole body manipulation motion planning is proposed. An object pickup scenario using the NASA-JSC Valkyrie robot is also provided in this paper to demonstrate the performance of the planning algorithm.

Human-in-the-loop control through kinematic redundancy resolution for space exploration rovers

We present an architecture to enable the modeling of human-in-the-loop control problem of space exploration robotic systems. We describe a shared control architecture, originally developed for assistive and disaster response robotics, for potential use in a space exploration scenario. Examples relevant to space are provided to highlight the different elements of the shared control architecture. A method for integrating human input with the autonomous behavior of the system using redundancy resolution taking into account primary and secondary tasks is presented. Five different modalities of shared control are described with the redundancy resolution method. The shared control architecture and redundancy resolution allow human input to be integrated as constraints on the kinematic model of the system.