Path planning and optimization play an important role in robotics and autonomous vehicles. In this work we address the particular problem of optimizing sets of non-overlapping paths which form lanes for the efficient flow of multiple agents in a given cluttered environment. The proposed method is based on a convex optimization formulation applied to a piecewise Quadratic Bezier path representation. Our method guarantees a given minimum clearance value while minimizing total length and maximum curvature. Clearance is addressed with respect to both obstacles and adjacent lanes, such that a non-overlapping set of lanes is obtained. The proposed method is based on interleaving the optimization of individual lanes according to a priority criterion that improves overall convergence. Users have the flexibility to customize the objective function by assigning different weights to the clearance, length, and curvature objective terms. We present results applied to sets of lanes generated by an RRT planner which has been extended to produce multiple initial lanes. Our results showcase the ability to produce different sets of lanes reflecting user-defined weights. We also present comparisons against a heuristic shortcut-based smoothing method and a generalized optimization formulation, which demonstrate the improved performance of the proposed approach.

We consider a local planner that utilizes model predictive control to locally deviate from a prescribed global path in response to dynamic environments, taking into account the system dynamics. To ensure the consistency between the local and global paths, we introduce the concept of locally homotopic paths for paths with different origins and destinations. We then formulate a hard constraint to ensure that local paths are locally homotopic to a given global path. Additionally, we propose a cost function to penalize any violation of this requirement, rather than completely prohibiting it. Experimental results show that both variants of our approach are more resilient to localization errors, compared to existing methods that represent the homotopy class constraint as an envelope around the global path.

The need to plan motions for agents with variable shape constraints such as under different formations appears in several virtual and real-world applications of autonomous agents. In this work, we focus on planning and execution of formation-aware paths for a group of agents traversing a cluttered environment. The proposed planning framework addresses the trade-off between being able to enforce a preferable formation when traversing the corridors of the environment, versus accepting to switch to alternative formations requiring less clearance in order to utilize narrower corridors that can lead to a shorter overall path to the final destination. At the planning stage, this trade-off is addressed with a multi-layer graph annotated with per-layer navigation costs and formation transition costs, where each layer represents one formation together with its specific clearance requirement. At the navigation stage, we introduce Corridor Shortest Path Maps (CSPMs), which produce a vector field for guiding agents along the solution corridor, ensuring unobstructed in-formation navigation in cluttered environments, as well as group motion along lengthwise-optimal paths in the solution corridor. We also present examples of how our multi-layer planning framework can be applied to other types of multi-modal planning problems. This work addresses multi-formation path planning and navigation in cluttered environments. A new multi-layer graph construction and search methodology is proposed for computing global solutions considering formation preferences and required clearances. At the navigation level we introduce Corridor Shortest Path Maps for achieving unobstructed in-formation group navigation.image

This article introduces the new problem of planning spatially distributed lanes for supporting multi‐agent navigation applications in 3D environments. Our proposed approach computes the max‐flow of a 3D medial axis representation of the environment in order to globally compute collision‐free lanes exploring the entire free space of the volumetric scene. Our method addresses agent clearance and path dispersion in order to provide a comprehensive solution to globally compute lanes to be used by multiple agents in 3D environments. By selecting the desired lane dispersion our approach offers an intuitive and powerful way to explore variations in the computed collections of lanes. Dispersion is addressed with a combination of new techniques based on max flow computation, clearance‐based path separation, and adaptive shortcut‐based smoothing.

Path optimization typically involves finding a smooth and short path that satisfies motion constraints of a given vehicle or simulated agent. In this paper, we propose a new path optimization scheme based on convex optimization of a piecewise quadratic Bézier path. Our method addresses $C^{1}$ continuity and a target minimum clearance to obstacles while minimizing path length and maximum curvature. The user is able to customize the objective function by assigning different weights to the clearance, length, and curvature terms. As a result our method is versatile and suitable for a variety of path planning applications in robotics and autonomous vehicles. We present several results demonstrating the ability to achieve paths according to different values for the curvature, length, and clearance parameters. Our benchmark scenarios apply our method directly to low-quality paths generated by a sampling-based planner, and as well in comparison to a heuristic shortcut-based smoothing method. Our results show that the quality of the paths produced by our method outperforms other approaches, and that our method is responsive to the weights chosen by the user. Our method is able to control the main path properties of interest in a unified fashion, and therefore represents an excellent option for several path planning applications in robotics and autonomous vehicles.

This paper proposes a methodology for computing, visualizing, and analyzing critical decision boundaries for the selection of shortest paths in a given environment. Decision boundaries are defined as the points in a map from which two or more different shortest paths exist towards a destination. This paper introduces the problem of visualizing their evolution, taking into account moving obstacles, moving goals, and as well multiple goals. The proposed visualizations enable analyzing which paths should be taken and at which departure times, such that a destination can be reached by the shortest possible path when taking into account a moving target or time-varying areas to be avoided. The proposed techniques are also applied to the analysis and improvement of exit placement in a given environment, in order to improve the evacuation flow in emergency situations.

In order to be successfully executed, collaborative tasks performed by two agents often require a cooperative strategy to be learned. In this work, we propose a constraint-based multi-agent reinforcement learning approach called Constrained Multi-agent Soft Actor Critic (C-MSAC) to train control policies for simulated agents performing collaborative multi-phase tasks. Given a task with n phases, the first n-1 phases are treated as constraints for the final task phase objective, which is addressed with a centralized training and decentralized execution approach. We highlight our framework on a tray balancing task including two phases: tray lifting and cooperative tray control for target following. We evaluate our proposed approach and compare it against its unconstrained variant (MSAC). The performed comparisons show that C-MSAC leads to higher success rates, more robust control policies, and better generalization performance.

Path smoothing is an important operation in a number of path planning applications. While several approaches have been proposed in the literature, a lack of simple and effective methods with quality-based termination conditions can be observed. In this paper we propose a deterministic shortcut-based smoothing method that is simple to be implemented and achieves user-specified termination conditions based on solution quality, overcoming one of the main limitations observed in traditional random-based approaches. We present several benchmarks demonstrating that our method produces higher-quality results when compared to the traditional random shortcuts approach.

We present a new methodology for enabling virtual humans to autonomously detect and perform complex full‐body interactions with their environments. Given a parameterized walking controller and a set of motion‐captured example interactions, our method is able to detect when interactions can occur and to coordinate the detected upper‐body interaction with the walking controller in order to achieve full‐body mobile interactions in similar situations. Our approach is based on learning spatial coordination features from the example motions and on associating body‐environment proximity information to the body configurations of each performed action. Body configurations become the input to a regression system, which in turn is able to generate new interactions for different situations in similar environments. The regression model is capable of selecting, encoding and replicating key spatial strategies with respect to body coordination and management of environment constraints as well as determining the correct moment in time and space for starting an interaction. As a result, we obtain an interactive controller able to detect and synthesize coordinated full‐body motions for a variety of complex interactions requiring body mobility. Our results achieve complex interactions, such as opening doors and drawing in a wide whiteboard. The presented approach introduces the concept of learning interaction coordination models that can be applied on top of any given walking controller. The obtained method is simple and flexible, it handles the detection of possible interactions and is suitable for real‐time applications.

One main challenge in multi-agent navigation is to generate trajectories minimizing bottlenecks in environments cluttered with obstacles. In this paper we approach this problem globally by taking into account the maximum flow capacity of a given polygonal environment. Given the difficulty in solving the continuous maximum flow of a planar environment, we present in this paper a GPU-based methodology which leads to practical methods for computing maximum flow maps in arbitrary two-dimensional polygonal domains. Once a flow map representation is obtained, lanes can be extracted and optimized in length while keeping constant the flow capacity achieved by the system of trajectories. This work extends our previous work on max flow maps by presenting a clearance-based flow generation method which takes into account the size of the agents at the flow generation phase. In this way we ensure that the maximum possible number of lanes with the needed clearance is always obtained, a property that was found to not be always obtained with our previous method. As a result we are able to generate trajectories of maximum flow from source to sink edges across a generic set of polygonal obstacles, enabling the deployment of large numbers of agents utilizing the maximum flow capacity of a continuous description of the environment and eliminating bottlenecks.

A navigation mesh is a representation of a 2D or 3D virtual environment that enables path planning and crowd simulation for walking characters. Various state-of-the-art navigation meshes exist, but there is no standardized way of evaluating or comparing them. Each implementation is in a different state of maturity, has been tested on different hardware, uses different example environments, and may have been designed with a different application in mind. In this paper, we develop and use a framework for comparing navigation meshes. First, we give general definitions of 2D and 3D environments and navigation meshes. Second, we propose theoretical properties by which navigation meshes can be classified. Third, we introduce metrics by which the quality of a navigation mesh implementation can be measured objectively. Fourth, we use these properties and metrics to compare various state-of-the-art navigation meshes in a range of 2D and 3D environments. Finally, we analyze our results to identify important topics for future research on navigation meshes. We expect that this work will set a new standard for the evaluation of navigation meshes, that it will help developers choose an appropriate navigation mesh for their application, and that it will steer future research in interesting directions. (C) 2020 Elsevier Ltd. All rights reserved.

Position-Based Dynamics (PBD) has been shown to provide a flexible framework for modeling per-agent collision avoidance behavior for crowd and multi-agent simulations in planar scenarios. In this work, we propose to extend the approach such that collision avoidance reactions can utilize in a controlled way the volumetric 3D space around each agent when deciding how to avoid collisions with other agents. We propose to use separation planes for collision avoidance, using either preferred or automatically determined planes. Our results demonstrate the ability to control the spatial 3D behavior of simulated agents by constraining the produced movements according to the separation planes. Our method is generic and can be integrated with different crowd simulation techniques. We also compare our results with a 3D collision avoidance method based on Reciprocal Velocity Obstacles (RVOs).

This paper describes a pipeline built with open source tools for interpolating 3D facial expressions taken from images. The presented approach allows anyone to create 3D face animations from 2 input photos: one from the start face expression, and the other from the final face expression. Given the input photos, corresponding 3D face models are constructed and texture-mapped using the photos as textures aligned with facial features. Animations are then generated by morphing the models by interpolation of the geometries and textures of the models. This work was performed as a MS project at the University of California, Merced.

In this paper we address feedback strategies for an autonomous virtual trainer. First, a pilot study was conducted to identify and specify feedback strategies for assisting participants in performing a given task. The task involved sorting virtual cubes according to areas of countries displayed on them. Two feedback strategies were specified. The first provides correctness feedback by fully correcting user responses at each stage of the task, and the second provides suggestive feedback by only notifying if and how a response can be corrected. Both strategies were implemented in a virtual training system and empirically evaluated. The correctness feedback strategy was preferred by the participants, was more effective time-wise, and was more effective in improving task performance skills. The overall system was also rated comparable to hypothetically performing the same task with real interactions.

The effective integration of local collision avoidance with global path planning becomes a necessity when multi-agent systems need to be simulated in complex cluttered environments. This work presents our first results exploring the new approach of integrating Shortest Path Maps (SPMs) with local collision avoidance in order to provide optimal paths for agents to navigate around obstacles toward their goal locations. Our GPU-based SPM implementation is available. CCS Concepts • Computing methodologies → Collision detection; Multi-agent planning;

This paper presents findings of a small-scale exploratory study that investigated effects of virtual agent gender on user performance and preference in a virtual reality (VR) training program. During the study, twelve participants, predominantly young male adults (10 males, average age 24 years), took part in a customized Box and Blocks Test (BBT), where virtual female and male agents instructed them to quickly move specific color cubes from one side of the table to the other side of the table. Although, on average, the male-dominated sample performed better with the male agent than the female agent, with respect to task completion time, error rate, and error correction time, no significant difference was identified. There was also no significant effect of agent gender on preference, perceived helpfulness, and perceived professionalism. However, interestingly, significantly more participants found the female agent more attractive than the male agent, presumably due to the male-dominated sample.

While significant research has been dedicated to the simulation of fluids, not much attention has been given to exploring new interesting behavior that can be generated with the different types of non-Newtonian fluids with non-constant viscosity. Going in this direction, this paper introduces a computational model for simulating the interesting phenomena observed in non-Newtonian shear thickening fluids, which are fluids where the viscosity increases with increased stress. These fluids have unique and unconventional behavior, and they often appear in real world scenarios such as when sinking in quicksand or when experimenting with popular cornstarch and water mixtures. While interesting behavior of shear thickening fluids can be easily observed in the real world, the most interesting phenomena of these fluids have not been simulated before in computer graphics. The fluid exhibits unique phase changes between solid and liquid states, great impact resistance in its solid state and strong hysteresis effects. Our proposed approach builds on existing non-Newtonian fluid models in computer graphics and introduces an efficient history-based stiffness term that is essential to produce the most interesting shear thickening phenomena. The history-based stiffness is formulated through the use of fractional derivatives, leveraging the fractional calculus ability to depict both the viscoelastic behavior and the history effects of history-dependent systems. Simulations produced by our method are compared against real experiments and the results demonstrate that the proposed model successfully captures key phenomena observed in shear thickening fluids.

We introduce a new method for computing optimal path maps on the GPU using OpenGL shaders. Our method explores GPU rasterization as a way to propagate optimal costs on a polygonal 2D environment, producing optimal path maps which can efficiently be queried at run-time. Our method is implemented entirely with GPU shaders, does not require pre-computation, addresses optimal path maps with multiple points and line segments as sources, and introduces a new optimal path map concept not addressed before: maps with weights at vertices representing possible changes in traversal speed. The produced maps offer new capabilities not explored by previous navigation representations and at the same time address paths with global optimality, a characteristic which has been mostly neglected in animated virtual environments. The proposed path maps partition the input environment into the regions sharing a same parent point along the shortest path to the closest source, taking into account possible speed changes at vertices. The proposed approach is particularly suitable for the animation of multiple agents moving toward the entrances or exits of a virtual environment, a situation which is efficiently represented with the proposed path maps.