ICRA 2021 Papers

Kinematics is the first consideration in designing the mechanical structures that comprise robots. Of the many subcategories that exist under this umbrella, an often early design goal is to achieve some desired workspace. This goal applies to both single and multi-degree-of-freedom systems. Previous literature has applied the diversity of extant optimization techniques for achieving such design goals. A conceptually simple approach to single-objective optimization is to symbolically derive a gradient vector, then find all of its zeroes. This approach is easier said than done since the resulting system is nonlinear. For this reason, sophisticated optimization heuristics are more commonly employed. In this paper, we revitalize the former approach, offering a route to efficiently find all of the gradient zeroes, including the global minimum. Our approach is facilitated by homotopy continuation. We connect the theoretical results to practical problems by demonstrating the design of a mechanism for a humanoid walking gait and the finger of a robotic hand.

The inspection-planning problem calls for computing motions for a robot that allow it to inspect a set of points of interest (POIs) while considering plan quality (e.g., plan length). This problem has applications across many domains where robots can help with inspection, including infrastructure maintenance, construction, and surgery. Incremental Random Inspection-roadmap Search (IRIS) is an asymptotically-optimal inspection planner that was shown to compute higher-quality inspection plans orders of magnitudes faster than the prior state-of-the-art method. In this paper, we significantly accelerate the performance of IRIS to broaden its applicability to more challenging real-world applications. A key computational challenge that IRIS faces is effectively searching roadmaps for inspection plans—a procedure that dominates its running time. In this work, we show how to incorporate lazy edge-evaluation techniques into IRIS’s search algorithm and how to reuse search efforts when a roadmap undergoes local changes. These enhancements, which do not compromise IRIS’s asymptotic optimality, enable us to compute inspection plans much faster than the original IRIS. We apply IRIS with the enhancements to simulated bridge inspection and surgical inspection tasks and show that our new algorithm for some scenarios can compute similar-quality inspection plans 570× faster than prior work.

This paper presents a computational framework for the design of high-performance legged robotic systems. The framework relies on the concurrent optimization of hardware parameters and control trajectories to find the best robot design for a given task. In particular, we focus on energy efficiency, presenting novel electro-mechanical models to account for the losses of the actuators due to friction and Joule effects. Thanks to a bi-level optimization scheme, featuring a genetic algorithm in the outer loop, our framework can also optimize for the duration of the motion, the actuators, and the size of the robot. We present a novel approach to scale both the actuators and the robot structure in a way that ensures structural integrity by maintaining constant the normalized deflection of the links. We validated our approach by designing a two-joint monoped robot to execute a jumping task. Our simulation results show that our framework can lead to remarkable energy savings (up to 60%) thanks to the concurrent optimization of robot size, motion duration, and actuators.

Orthogonally assembled double-layered corrugated (OADLC) mechanisms are a class of foldable structures that harness origami-inspired methods to enhance the structural stiffness of resulting devices; these mechanisms have extensive applications due to their lightweight, compact nature as well as their high strength-to-weight ratio. However, the design of these mechanisms remains challenging. Here, we propose an efficient method to rapidly design OADLC mechanisms from desired behavioral specifications, i.e. in-plane stiffness and out-of-plane stiffness. Based on an equivalent plate model, we develop and validate analytical formulas for the behavioral specifications of OADLC mechanisms; the analytical formulas can be described as expressions of design parameters. On the basis of the analytical expressions, we formulate the design of OADLC mechanisms from behavioral specifications into an optimization problem that minimizes the weight with given design constraints. The 2D folding patterns of the optimized OADLC mechanisms can be generated automatically and directly delivered for fabrication. Our rapid design method is demonstrated by developing stiffness-enhanced mechanisms with a desired out-of-plane stiffness for a foldable gripper that enables a blimp to perch steadily under air disturbance and weight limit.

We present a fast, scalable, and accurate Simultaneous Localization and Mapping (SLAM) system that represents indoor scenes as a graph of objects. Leveraging the observation that artificial environments are structured and occupied by recognizable objects, we show that a compositional and scalable object mapping formulation is amenable to a robust SLAM solution for drift-free large-scale indoor reconstruction. To achieve this, we propose a novel semantically assisted data association strategy that results in unambiguous persistent object landmarks and a 2.5D compositional rendering method that enables reliable frame-to-model RGB-D tracking. Consequently, we deliver an optimized online implementation that can run at near frame rate with a single graphics card, and provide a comprehensive evaluation against state-of-the-art baselines. An open-source implementation will be provided at https://github.com/rpl-cmu/object-slam.

The research and development of socially interactive robots is a complex challenge because of the wide variety of capabilities needed for effective social human-robot interactions (HRI). Many of these capabilities, including perception, dialog, and control, have state of the art methods and solutions, but combining those into a comprehensive and seamless interaction is still an open challenge. We describe HARMONI, a multi-modal, open-source tool for rapid social HRI development and deployment. HARMONI is centered around a ROS package for interaction development, including decision management and node orchestration. HARMONI systematically integrates with disparate functionalities needed to conduct a meaningful social human-robot interaction such as external cloud services, AI models, and modules for sensing, planning, and acting on a variety of platforms. HARMONI was applied to the QT robot platform and usability tests were conducted to evaluate the ease and speed of development and deployment. This paper describes the architecture and design of HARMONI and reports the results of a pilot study with novice users.

Despite decades of work in fast reactive planning and control, challenges remain in developing reactive motion policies on non-Euclidean manifolds and enforcing constraints while avoiding undesirable potential function local minima. This work presents a principled method for designing and fusing desired robot task behaviors into a stable robot motion policy, leveraging the geometric structure of non-Euclidean manifolds, which are prevalent in robot configuration and task spaces. Our Pullback Bundle Dynamical Systems (PBDS) framework drives desired task behaviors and prioritizes tasks using separate position-dependent and position/velocity-dependent Riemannian metrics, respectively, thus simplifying individual task design and modular composition of tasks. For enforcing constraints, we provide a class of metric-based tasks, eliminating local minima by imposing non-conflicting potential functions only for goal region attraction. We also provide a geometric optimization problem for combining tasks inspired by Riemannian Motion Policies (RMPs) that reduces to a simple least-squares problem, and we show that our approach is geometrically well-defined. We demonstrate the PBDS framework on the sphere S2 and at 300-500 Hz on a manipulator arm, and we provide task design guidance and an open-source Julia library implementation. Overall, this work presents a fast, easy-to-use framework for generating motion policies without unwanted potential function local minima on general manifolds.

This paper considers a fundamental, optimal path planning problem that requires simultaneously minimizing path length and maximizing obstacle clearance. We show that in even simple planar settings with point and disc obstacles, the set of alternative solutions such that no one is clearly better than another (the set of Pareto-optimal solutions) is uncountably infinite. In spite of this difficulty, we introduce a complete, efficient algorithm that computes the Pareto front and a data structure that finitely represents the complete set of all Pareto- optimal paths. Particular optimal paths can then be selected from the computed data structure during execution, based on any additional conditions or considerations.

Soft manipulators made of deformable materials have great promise in applications that require additional flexibility and compliance; however, these characteristics also make them difficult to simulate accurately and quickly. The lack of a fast and accurate simulator prevents motion planners from generating feasible plans, which would enable soft robots to achieve more complex tasks, such as manipulation. In this work, we propose combining a simplified quasistatic model with a neural network that learns to compensate for unmodeled forces, such as friction and loads, in order to create a fast forward model for soft manipulators with multiple segments. We show that the resulting neural network model reduces average end effector position error by 62% compared to the quasistatic model, while still being fast enough for motion planning. We also incorporate this model into an RRT*-based planner and demonstrate that the plans generated using our model are more likely to be feasible when executed on hardware than plans generated with a simulator using the quasistatic model.

This paper introduces the Compartmentalized Covariance Intersection (CCI) algorithm, a consistent technique to fuse measurements in cooperative navigation networks. The algorithm reduces the excess conservatism of standard Covariance Intersection (CI) by assuming that correlation is only present within each measurement stream and not across the different sources. This assumption allows the sources to be compartmentalized and fused with the Kalman equations rather than the CI method, resulting in tighter convergence. The CCI algorithm is applied to a cooperative localization application and is demonstrated to substantially outperform CI, with a covariance that approaches the performance of an ideal centralized estimator in simulations of linear, Gaussian systems. This approach is also demonstrated to be consistent using Monte Carlo simulations. Finally, CCI was used in a laboratory demonstration to perform distributed localization with real-world range measurements between a team of six agents.

As locomotion decisions must be taken by considering the future, most existing quadruped controllers are based on a model predictive controller (MPC) with a reduced model of the dynamics to generate the motion and a whole- body controller to execute it. Yet the simplifying assumptions of the MPC are often chosen ad-hoc or by intuition. In this article, we focus on a set of MPCs and analyze the effect of chosen model reductions on the behavior of the robot. Based on existing formulations, we present additional controllers to better understand the influence of model reductions on the controller capabilities. Finally, we propose a robust predictive controller capable of optimizing the foot placements, gait period, center- of-mass trajectory and ground reaction forces. The behavior of these controllers is statistically evaluated in simulation. This empirical study aims to assess the relative importance of the components of the optimal control problem (variables, costs, dynamics) to be able to take reasoned decisions instead of arbitrarily emphasizing or neglecting some of them. We also provide a qualitative study in simulation and on the real robot Solo-12.

Submodular maximization has been widely used in many multi-robot task planning problems including information gathering, exploration, and target tracking. However, the interplay between submodular maximization and communication is rarely explored in the multi-robot setting. In many cases, maximizing the submodular objective may drive the robots in a way so as to disconnect the communication network. Driven by such observations, in this paper, we consider the problem of maximizing submodular function with connectivity constraints. Specifically, we propose a problem called Communication-aware Submodular Maximization (CSM), in which communication maintenance and submodular maximization are jointly considered in the decision-making process. One heuristic algorithm that consists of two stages, i.e. topology generation and deviation minimization is proposed. We validate the formulation and algorithm through numerical simulation. We find that our algorithm on average suffers only slightly performance decrease compared to the pure greedy strategy.

Service robots are gaining capabilities to be deployed in public environments for human assistance. While robot actively providing guidance has shown great success in field study, the communication strategy (the strategy to decide whom to initiate the service for and when), and hence the performance evaluation, has been based on behavioral-based qualitative analysis. We attribute this to the challenge of accessing large-scale field data with condition control, and approach the problem with simulation from the agent-based modeling literature, to simulate pedestrian behavior in unfamiliar environments and estimate travel cost. We contribute a planning approach that uses the pedestrian behavior prediction from the model, to decide whom to initiate guidance and when for performance maximization. The results suggest that our approach is more efficient based on the measure of saved pedestrian travel time, compared to the behavioral-based strategy and a baseline that maximizes service counts.

This paper investigates the tracking control problem of a class of high-order distributed systems subjected to limited communication bandwidth. An event-triggered control method is proposed, where the controller is triggered only when specific events happen. Moreover, the computational complexity is reduced by introducing command filters for virtual control signals. Specifically, the backstepping scheme is adopted as the main design framework, by which the n-th order nonlinear system is divided into n command-cascaded first order subsystems. And virtual control commands are sent through a second-order low-pass filter, by which the time derivatives of the virtual commands can be obtained directly. The theoretical analysis shows the stability of the proposed method. The tracking performance is illustrated by a simulation example.

We propose a combination of Model Predictive Control and Lexicographic Programming to address complex scenarios with conflicting goals related to various aspects of comfort and safety of passengers in a transit bus, generating different deceleration profiles depending on the speed of the bus and distance to obstacles, validated in experiments with a standard transit bus equipped with self-driving capabilities.

This paper addresses the problem of automating fleets of robots with non-holonomic dynamics. Previously studied methods either specialize in facets of this problem, that is, one or a combination of multi-robot goal allocation, motion planning, and coordination, and typically acrifice optimality and completeness for scalability. We propose an approach that constructs an abstract multi-robot roadmap in a reduced configuration space, where we account for environment connectivity and interference cost between robots occupying the same polygons. Querying the road-map results in a robot-goal assignment and abstract multi-robot trajectory. This is then exploited to de-compose the original problem into smaller problems, each of which is solved with a multi-robot motion planner that accounts for kinodynamic constraints. We validate the approach experimentally to demonstrate the advantage of considering task assignment and motion planning holistically, and explore some methods for balancing solution quality and computational efficiency.

Precise system identification is an important aspect of adequate control design and parameter definition to allow for accurate and reliable navigation. While this is well known in robotics, the community working with small rotorcraft Unmanned Aerial Vehicles (UAVs) has yet to discover the benefits. In contrast to existing work, which often performs offline or deterministic (i.e. closed-form) system identification, we present a probabilistic approach to the online estimation of system identification parameters and self-calibration states. Instead of decoupling system identification and state estimation for vehicle control, we merge the entire process into a holistic probabilistic framework to allow self-awareness and self-healing. Our observability analysis shows that most of the system identification parameters are observable and converge quickly to the optimal value using a combination of inertial cues, dynamic modeling, and an additional exteroceptive sensor. We support our theoretical findings with extensive tests simulating realistic data in Gazebo.

Planning for legged-wheeled machines is typically done using trajectory optimization because of many degrees of freedom, thus rendering legged-wheeled planners prone to falling prey to bad local minima. We present a combined sampling and optimization-based planning approach that can cope with challenging terrain. The sampling-based stage computes whole-body configurations and contact schedule, which speeds up the optimization convergence. The optimization-based stage ensures that all the system constraints, such as non-holonomic rolling constraints, are satisfied. The evaluations show the importance of good initial guesses for optimization. Furthermore, they suggest that terrain/collision (avoidance) constraints are more challenging than the robot model’s constraints. Lastly, we extend the optimization to handle general terrain representations in the form of elevation maps.

In this work, we present a per-instant pose optimization method that can generate configurations that achieve specified pose or motion objectives as best as possible over a sequence of solutions, while also simultaneously avoiding collisions with static or dynamic obstacles in the environment. We cast our method as a weighted sum non-linear constrained optimization-based IK problem where each term in the objective function encodes a particular pose objective. We demonstrate how to effectively incorporate environment collision avoidance as a single term in this multi-objective, optimization-based IK structure, and provide solutions for how to spatially represent and organize external environments such that data can be efficiently passed to a real-time, performance-critical optimization loop. We demonstrate the effectiveness of our method by comparing it to various state-of-the-art methods in a testbed of simulation experiments and discuss the implications of our work based on our results.