Dylan A. Shell

The role of physical embodiment in human-robot interaction

Autonomous robots are agents with physical bodies that share our environment. In this work, we test the hypothesis that physical embodiment has a measurable effect on performance and perception of social interactions. Support of this hypothesis would suggest fundamental differences between virtual agents and robots from a social standpoint and have significant implications for human-robot interaction. We measure task performance and perception of a robots social abilities in a structured but open-ended task based on the Towers of Hanoi puzzle. Our experiment compares aspects of embodiment by evaluating: (1) the difference between a physical robot and a simulated one; (2) the effect of physical presence through a co-located robot versus a remote tele-present robot. We present data from a pilot study with 12 subjects showing interesting differences in perception of remote physical robots and simulated agents attention to the task, and task enjoyment

On foraging strategies for large-scale multi-robot systems

Physical interference limits the utility of large-scale multi-robot systems. We present an empirical study of the effects of such interference in systems with hundreds of minimalist robots. We consider the canonical multi-robot foraging task, and define a new parametrized controller. This controller allows for evaluation of spatial arbitration strategies along a continuum with the traditional homogeneous and bucket-brigading algorithms at each end. We present data from thousands of simulations which suggests that methods surprisingly close to homogeneous foraging, but augmented with limited arbitration, can improve both performance and reliability

Large-scale multi-robot task allocation via dynamic partitioning and distribution

This paper introduces an approach that scales assignment algorithms to large numbers of robots and tasks. It is especially suitable for dynamic task allocations since both task locality and sparsity can be effectively exploited. We observe that an assignment can be computed through coarsening and partitioning operations on the standard utility matrix via a set of mature partitioning techniques and programs. The algorithm mixes centralized and decentralized approaches dynamically at different scales to produce a fast, robust method that is accurate and scalable, and reduces both the global communication and unnecessary repeated computation. An allocation results by operating on each partition: either the steps are repeated recursively to refine the generalized assignment, or each sub-problem may be solved by an existing algorithm. The results suggest that only a minor sacrifice in solution quality is needed for significant gains in efficiency. The algorithm is validated using extensive simulation experiments and the results show advantages over the traditional optimal assignment algorithms.

Extending open dynamics engine for robotics simulation

Open Dynamics Engine (ODE) is the most popular rigidbody dynamics implementation for robotics simulation applications. While using it to simulate common robotic scenarios like mobile robot locomotion and simple grasping, we have identified the following shortcomings each of which adversely affect robot simulation: lack of computational efficiency, poor support for practical joint-dampening, inadequate solver robustness, and friction approximation via linearization. In this paper we describe extensions to ODE that address each of these problems. Because some of these objectives lie in opposition to others--e.g., speed versus verisimilitude--we have carried out experiments in order to identify the trade-offs involved in selecting from our extensions. Results from both elementary physics and robotic task-based scenarios show that speed improvements can be gained along with useful joint-dampening. If one is willing to accept an execution time cost, we are able to represent the full-friction cone, while simultaneously guaranteeing a solution from our numerical solver.

Assessing optimal assignment under uncertainty: An interval-based algorithm

We consider the problem of multi-robot task-allocation when robots have to deal with uncertain utility estimates. Typically an allocation is performed to maximize expected utility; we consider a means for measuring the robustness of a given optimal allocation when robots have some measure of the uncertainty (e.g. a probability distribution, or moments of such distributions). We introduce the interval Hungarian algorithm, a new algorithm that extends the classic Kuhn—Munkres Hungarian algorithm to compute the maximum interval of deviation, for each entry in the assignment matrix, which will retain the same optimal assignment. The algorithm has a worst-case time complexity of O(n 4); we also introduce a parallel variant with O(n 3) running time, which is able to exploit the concurrent computing capabilities of distributed multi-robot systems. This provides an efficient measurement of the tolerance of the allocation to the uncertainties and dynamics, for both a specific interval and a set of interrelated intervals. We conduct experiments both in simulation and with physical robots to validate the approach and to gain insight into the effect of location uncertainty on allocations for multi-robot multi-target navigation tasks.

Modeling Contact Friction and Joint Friction in Dynamic Robotic Simulation Using the Principle of Maximum Dissipation

We present a unified treatment for modeling Coulomb and viscous friction within multi-rigid body simulation using the principle of maximum dissipation. This principle is used to build two different methods—an event-driven impulse-based method and a time stepping method—for modeling contact. The same principle is used to effect joint friction in articulated mechanisms. Experiments show that the contact models are able to be solved faster and more robustly than alternative models. Experiments on the joint friction model show that it is as accurate as a standard model while permitting much larger simulation step sizes to be employed.

Multi-Level Partitioning and Distribution of the Assignment Problem for Large-Scale Multi-Robot Task Allocation.

A team of robots can handle failures and dynamic tasks by repeatedly assigning functioning robots to tasks. This paper introduces an algorithm that scales to large numbers of robots and tasks by exploiting both task locality and sparsity. The algorithm mixes both centralized and decentralized approaches at different scales to produce a fast, robust method that is accurate and scalable, and reduces both the global communication and unnecessary repeated computation. We depart from optimization and bipartite matching formulations of the problem, observing instead that an assignment can be computed through coarsening and partitioning operations on the utility matrix. First, a coarse assignment is calculated by evaluating the global utility information and partitioning it into clusters in a problem-domain independent way. Next, the assignment solutions in each partition are refined (either recursively, or via an existing algorithm). This multilevel framework allows the repeated reassignment to execute among interrelated partitions. The results suggest that only a minor sacrifice in solution quality is required for gains in efficiency. The proposed algorithm is validated using extensive simulation experiments and the results show advantages over the traditional optimal assignment algorithms.

Insights Toward Robot-Assisted Evacuation

This paper considers the problem of human evacuation assistance. We discuss how this kind of task differs from the more prevalent search and rescue tasks, and the resulting implications for the design of assistive evacuation systems. We describe the implementation and evaluation of an algorithm for deploying audio navigational cues throughout an office building with a team of mobile robots. A review of evacuation dynamics methods is presented and particular methods are applied on-line during deployment. We use a pedestrian simulation and a simple model of audio–evacuee interaction to show the effects of beacon deployment; the results indicate that even a small number of beacons can significantly decrease the mean and variance of egress time and distance. We also advance uses for human motion-based measures of environment complexity for general mobile robotics.

A Distributable and Computation-flexible Assignment Algorithm: From Local Task Swapping to Global Optimality

The assignment problem arises in multi-robot taskallocation scenarios. This paper introduces an algorithm for solving the assignment problem with several appealing features for online, distributed robotics applications. The method can start with any initial matching and incrementally improve the solution to reach the global optimum, producing valid assignments at any intermediate point. It is an any-time algorithm with an attractive performance profile (quality improves linearly) that, additionally, is comparatively straightforward to implement and is efficient both theoretically (O(n lgn) complexity is better than widely used solvers) and practically (comparable to the fastest implementation, for up to hundreds of robots/tasks). We present a centralized version and two decentralized variants that trade between computational and communication complexity. Inspired by techniques that employ task exchanges between robots, our algorithm guarantees global optimality while using generalized “swap” primitives. The centralized version turns out to be a computational improvement and reinterpretation of the little-known method of Balinski-Gomory, proposed half a century ago. Deeper understanding of the relationship between approximate swap-based techniques —developed by roboticists— and combinatorial optimization techniques, e.g., the Hungarian and Auction algorithms —developed by operations researchers but used extensively by roboticists— is uncovered.

Optimal Market-based Multi-Robot Task Allocation via Strategic Pricing.

Auction and market-based mechanisms are among the most popular methods for distributed task allocation in multirobot systems. Most of these mechanisms were designed in a heuristic way and analysis of the quality of the resulting assignment solution is rare. This paper presents a new market-based multi-robot task allocation algorithm that produces optimal assignments. Rather than adopting a buyer’s “selfish” bidding perspective as in previous auction/market-based approaches, the proposed method approaches auctioning from a merchant’s point of view, producing a pricing policy that responds to cliques of customers. The algorithm uses price escalation to clear a market of all its goods, producing a state of equilibrium that satisfies both the merchant and customers. The proposed method can be used as a general assignment algorithm as it has a time complexity (O(nlgn)) close to the fastest state-of-the-art algorithms (O(n)) but is extremely easy to implement. As in previous research, the economic model reflects the distributed nature of markets inherently: in this paper it leads directly to a decentralized method ideally suited for distributed multi-robot systems.