Sisir Karumanchi

An Architecture for Online Affordance-based Perception and Whole-body Planning

The DARPA Robotics Challenge Trials held in December 2013 provided a landmark demonstration of dexterous mobile robots executing a variety of tasks aided by a remote human operator using only data from the robots sensor suite transmitted over a constrained, field-realistic communications link. We describe the design considerations, architecture, implementation, and performance of the software that Team MIT developed to command and control an Atlas humanoid robot. Our design emphasized human interaction with an efficient motion planner, where operators expressed desired robot actions in terms of affordances fit using perception and manipulated in a custom user interface. We highlight several important lessons we learned while developing our system on a highly compressed schedule.

Constraint-based planning and control for safe, semi-autonomous operation of vehicles

This paper presents a new approach to semi-autonomous vehicle hazard avoidance and stability control, based on the design and selective enforcement of constraints. This differs from traditional approaches that rely on the planning and tracking of paths. This emphasis on constraints facilitates “minimally-invasive” control for human-machine systems; instead of forcing a human operator to follow an automation-determined path, the constraint-based approach identifies safe homotopies, and allows the operator to navigate freely within them, introducing control action only as necessary to ensure that the vehicle does not violate safety constraints. The method evaluates candidate homotopies based on “restrictiveness”, rather than traditional measures of path goodness, and designs and enforces requisite constraints on the humans control commands to ensure that the vehicle never leaves the controllable subset of a desired homotopy. Identification of these homotopic classes in off-road environments is performed using geometric constructs. The goodness of competing homotopies and their associated constraints is then characterized using geometric heuristics. Finally, input limits satisfying homotopy and vehicle dynamic constraints are enforced using threat-based feedback mechanisms to ensure that the vehicle avoids collisions and instability while preserving the human operators situational awareness and mental models. The methods developed in this work are shown in simulation and experimentally demonstrated in safe, high-speed teleoperation of an unmanned ground vehicle.

Non-parametric Learning to Aid Path Planning over Slopes

In this paper we address the problem of closing the loop from perception to action selection for unmanned ground vehicles, with a focus on navigating slopes. A new non-parametric learning technique is presented to generate a mobility representation where the maximum feasible speed is used as a criterion to classify the world. The inputs to the algorithm are terrain gradients derived from an elevation map and past observations of wheel slip. It is argued that such a representation can aid in path planning with improved selection of vehicle heading and velocity in off-road slopes. In addition, an information theoretic test is proposed to validate a chosen proprioceptive representation (such as slip) for mobility map generation. Results of mobility map generation and its benefits to path planning are shown.

Supervised Remote Robot with Guided Autonomy and Teleoperation (SURROGATE): A framework for whole-body manipulation

The use of the cognitive capabilties of humans to help guide the autonomy of robotics platforms in what is typically called “supervised-autonomy” is becoming more commonplace in robotics research. The work discussed in this paper presents an approach to a human-in-the-loop mode of robot operation that integrates high level human cognition and commanding with the intelligence and processing power of autonomous systems. Our framework for a “Supervised Remote Robot with Guided Autonomy and Teleoperation” (SURROGATE) is demonstrated on a robotic platform consisting of a pan-tilt perception head, two 7-DOF arms connected by a single 7-DOF torso, mounted on a tracked-wheel base. We present an architecture that allows high-level supervisory commands and intents to be specified by a user that are then interpreted by the robotic system to perform whole body manipulation tasks autonomously. We use a concept of “behaviors” to chain together sequences of “actions” for the robot to perform which is then executed real time.

Homotopy-Based Divide-and-Conquer Strategy for Optimal Trajectory Planning via Mixed-Integer Programming

This paper proposes an optimal trajectory generation framework in which the global obstacle-avoidance problem is decomposed into simpler subproblems, corresponding to distinct path homotopies. In classical approaches to homotopic trajectory planning, trajectory planning and homotopy identification are performed simultaneously, leading to a substantial computational burden. The main benefit of the proposed approach is the development of a method to enumerate and explicitly represent distinct homotopy classes before trajectory planning or optimization, which allow the problem to be decomposed into simpler independent subproblems. The main contribution of the paper is twofold. The first contribution is the description of a method for utilizing existing cell-decomposition methods to enumerate and represent local trajectory generation problems that can be solved efficiently and independently. In addition, a relationship between the proposed cell-sequence representation and homotopy classes is analyzed. The second contribution is a computationally efficient novel formulation of the trajectory optimization problem within a cell sequence via mixed-integer quadratic programming (MIQP). Computational efficiency and increased solution richness of the proposed approach are demonstrated through simulation studies. The proposed MIQP formulation fits into a linear model-predictive control framework with nonconvex collision-free constraints.

Reactive control in environments with hard and soft hazards

In this paper we present a generalization of reactive obstacle avoidance algorithms for mobile robots operating among soft hazards such as off-road slopes and deformable terrain. A new hazard avoidance scheme generalizes constraint based reactive algorithms [1], [2] from hard to soft hazards. Reactive controllers operate by directly parameterizing the closedloop dynamics of the system with respect to the environment the robot is operating in. Traditionally, reactive controllers are parameterized by weighting virtual attraction and repulsion forces from goals and obstacles [3], [4]. One pitfall of such parameterizations is sensitivity of the tuning parameters to the operating environment. A reactive controller tuned in one set of conditions is not applicable in another (e.g. a different density of obstacles). The algorithm presented in this paper has two key properties which are significant i) Parameterization is environment independent. ii) It can deal with non-binary environments that contain soft hazards.

Modeling and traversal of pliable materials for tracked robot navigation

In order to fully exploit robot motion capabilities in complex environments, robots need to reason about obstacles in a non-binary fashion. In this paper, we focus on the modeling and characterization of pliable materials such as tall vegetation. These materials are of interest because they are pervasive in the real world, requiring the robotic vehicle to determine when to traverse or avoid them. This paper develops and experimentally verifies a template model for vegetation stems. In addition, it presents a methodology to generate predictions of the associated energetic cost incurred by a tracked mobile robot when traversing a vegetation patch of variable density.