Real-world environments are inherently uncertain, and to operate safely in these environments robots must be able to plan around this uncertainty. In the context of motion planning, we desire systems that can maintain an acceptable level of safety as the robot moves, even when the exact locations of nearby obstacles are not known. In this paper, we solve this chance-constrained motion planning problem using a sequential convex optimization framework. To constrain the risk of collision incurred by planned movements, we employ geometric objects called ε-shadows to compute upper bounds on the risk of collision between the robot and uncertain obstacles. We use these ε-shadow-based estimates as constraints in a nonlinear trajectory optimization problem, which we then solve by iteratively linearizing the non-convex risk constraints. This sequential optimization approach quickly finds trajectories that accomplish the desired motion while maintaining a user-specified limit on collision risk. Our method can be applied to robots and environments with arbitrary convex geometry; even in complex environments, it runs in less than a second and provides provable guarantees on the safety of planned trajectories, enabling fast, reactive, and safe robot motion in realistic environments.

Humans perceive and describe their surroundings with qualitative statements (e.g., "Alice's hand is in contact with a bottle."), rather than quantitative values (e.g., 6-D poses of Alice's hand and a bottle). Qualitative spatial representation (QSR) is a framework that represents the spatial information of objects in a qualitative manner. Region connection calculus (RCC), qualitative trajectory calculus (QTC), and qualitative distance calculus (QDC) are some popular QSR calculi. With the recent development of computer vision, it is important to compute QSR calculi from the visual inputs (e.g., RGB-D images). In fact, many QSR application domains (e.g., human activity recognition (HAR) in robotics) involve visual inputs. We propose a qualitative spatial representation network (QSRNet) that computes the three QSR calculi (i.e., RCC, QTC, and QDC) from the RGB-D images. QSRNet has the following novel contributions. First, QSRNet models the dependencies among the three QSR calculi. We introduce the dependencies as kinematics for QSR because they are analogous to the kinematics in classical mechanics. Second, QSRNet applies the 3-D point cloud instance segmentation to compute the QSR calculi. The experimental results show that QSRNet improves the accuracy in comparison to the other state-of-the-art techniques.

This paper introduces Probabilistic Chekov (p-Chekov), a chance-constrained motion planning system that can be applied to high degree-of-freedom (DOF) robots under motion uncertainty and imperfect state information. Given process and observation noise models, it can find feasible trajectories which satisfy a user-specified bound over the probability of collision. Leveraging our previous work in deterministic motion planning which integrated trajectory optimization into a sparse roadmap framework, p-Chekov shows superiority in its planning speed for high-dimensional tasks. P-Chekov incorporates a linear-quadratic Gaussian motion planning approach into the estimation of the robot state probability distribution, applies quadrature theories to waypoint collision risk estimation, and adapts risk allocation approaches to assign allowable probabilities of failure among waypoints. Unlike other existing risk-aware planners, p-Chekov can be applied to high-DOF robotic planning tasks without the convexification of the environment. The experiment results in this paper show that this p-Chekov system can effectively reduce collision risk and satisfy user-specified chance constraints in typical real-world planning scenarios for high-DOF robots.

Deployment of motion planning algorithms in practical applications has lagged due to their slow speed in reacting to disturbances. We believe that the best way to address this is to reuse learned planning and control information across queries. In previous work, we introduced Chekov, a reactive, integrated motion planning and execution system that reuses learned information in the form of an enhanced roadmap. We have previously shown how we can use Chekov to formulate trajectory optimization problems that result in superior performance in static environments. In this work, we show how incremental planning can be incorporated into the formulation of optimized trajectories from roadmap seed trajectories. Further, we show how an incremental planner can be adapted to reduce the overhead incurred for replanning when trajectories become invalid during execution.

Human activity recognition is a crucial ingredient in safe and efficient human-robot collaboration. In this paper, we present a new model-based human activity recognition system called logical activity recognition system (LCARS). LCARS requires much less training data compared to learning-based works. Compared to other model-based works, LCARS requires minimal domain-specific modeling effort from users. The minimal modeling is for two reasons: i) we provide a systematic and intuitive way to encode domain knowledge for LCARS and ii) LCARS automatically constructs a probabilistic estimation model from the domain knowledge. Requiring minimal training data and modeling effort allows LCARS to be easily applicable to various scenarios. We verify this through simulations and experiments.

We present an evaluation of several representative sampling-based and optimization-based motion planners, and then introduce an integrated motion planning system which incorporates recent advances in trajectory optimization into a sparse roadmap framework. Through experiments in 4 common application scenarios with 5000 test cases each, we show that optimization-based or sampling-based planners alone are not effective for realistic problems where fast planning times are required. To the best of our knowledge, this is the first work that presents such a systematic and comprehensive evaluation of state-of-the-art motion planners, which are based on a significant amount of experiments. We then combine different stand-alone planners with trajectory optimization. The results show that the combination of our sparse roadmap and trajectory optimization provides superior performance over other standard sampling-based planners' combinations. By using a multi-query roadmap instead of generating completely new trajectories for each planning problem, our approach allows for extensions such as persistent control policy information associated with a trajectory across planning problems. Also, the sub-optimality resulting from the sparsity of roadmap, as well as the unexpected disturbances from the environment, can both be overcome by the real-time trajectory optimization process.

Recent humanoid control investigations have emphasized the importance of controlling whole-body angular momentum throughout a movement task. For typical movement tasks, such as normal walking, such controllers minimize fluctuations in angular momentum about the center of mass (CM). This minimization is consistent with observed behavior of humans for such tasks. However, there are cases where such minimization is not desirable. In this study, we investigate movement tasks where bipedal balance control requires a relaxation of the goal of minimizing whole-body angular moment. We construct a humanoid model having a human-like mass distribution, and a Moment-Exploiting Control algorithm that modulates whole-body angular momentum to enhance CM control. The model only requires reference trajectories for CM position and torso orientation. Joint reference trajectories are not required. While balancing on one leg, we show that the controller is capable of correcting errors in CM state by sacrificing angular postural goals for the swing leg, trunk and head. This prioritization capability provides robustness to significant disturbances, without the need to plan new reference trajectories. We compare the dynamic behavior of our humanoid model to that of human test participants. While standing on one leg, the model, like the human, is shown to reposition its CM just above the stance foot from an initial body state where CM velocity is zero, and the ground CM projection falls outside the foot envelope.

A primary motivation for employing quadrupedal robots is that their morphology allows them to traverse difficult terrain. For example, a mountain goat, by carefully choosing its foot placements, is able to scale steep cliff sides. In contrast, wheeled robots have difficulty traveling over non-level terrain, and bipedal robots face stability challenges on rough terrain, even at low velocities. In order for quadrupeds to perform traversals over rough terrain in a stable manner, robust navigation strategies are needed that allow the robots to take full advantage of their physical capabilities. Foot placement and body pose planning is one of the most challenging problems associated with such navigation. We approach this problem as a combinatoric search over candidate foot placements and body poses. The search returns the sequence of kinematically feasible steps with the lowest cost as determined by their deviation from the terrain-independent nominal steps. Due to the large search domain in this problem and the speed required by real time robots, searching for the true optimal solution is computationally intractable. Therefore, we use a limited-horizon best-first search that quickly finds a near-optimal feasible solution. We show, through a series of tests, that this algorithm is sufficient for traversing challenging terrain, with obstacle heights approaching the leg length of the quadruped.

Motivated by biomechanical studies on human walking, we present a control strategy for biologically realistic walking based on the principle of spin angular momentum regulation. Using a morphologically realistic human model and kinematic gait data, we compute the total spin angular momentum at a self-selected walking speed for one human test subject. We find that dimensionless spin angular momentum remains small (S/sub 1//(mass height velocity) < 0.02) throughout the gait cycle, and maximum whole body angular excursions within sagittal (<1/spl deg/), coronal (<0.2/spl deg/), and transverse (<2/spl deg/) planes are negligible. These data support the hypothesis that spin angular momentum in human walking is highly regulated by the central nervous system, and that there exists a nonlinear coupling between ground reaction force, F~, center of mass position, r~/sub CM/ , and center of pressure location, r~/sub CP/, or F~ = (FZ/sub ///Z/sub CM/)(r~/sub CM/ -r~/sub CP/). We employ this relationship to rapidly generate biologically realistic CP and CM reference trajectories. Using an open loop optimization strategy, we show that biologically realistic leg joint kinematics emerge through the minimization of spin angular momentum and the total sum of joint torque squared, suggesting that both angular momentum and energetic factors are important considerations for biomimetic controllers.

We present an algorithm that provides enhanced flexibility and robustness in the control of single-leg humanoid standing through the coordination of stance leg ankle torques and stabilizing movements of non-contact limbs. Current control approaches generally assume the presence of explicitly specified joint reference trajectories or desired virtual force calculations that ignore system dynamics. Here we describe a practical controller that 1) simplifies control of abstract variables such as the center of mass location using a two-stage model-based plant linearization; 2) determines motion of non-contact limbs useful for achieving control targets while satisfying dynamic balance constraints; and 3) provides robustness to modeling error using a sliding controller. The controller is tested with a morphologically realistic, 3-dimensional, 18 degree-of-freedom humanoid model serving as the plant. It is demonstrated that the controller can use less detailed control targets, and reject stronger disturbances, than previously implemented controllers that employ desired virtual forces and static body calculations.