Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

Mexico City : México

The fast and cost-efficient home delivery of goods ordered online is logistically challenging. Many companies are looking for new ways to cross the last mile to their customers. One technology-enabled opportunity that recently has received much attention is the use of drones to support deliveries. An innovative last-mile delivery concept in which a truck collaborates with a drone to make deliveries gives rise to a new variant of the traveling salesman problem TSP that we call the TSP with drone. In this paper, we model this problem as an integer program and develop several fast route-first, cluster-second heuristics based on local search and dynamic programming. We prove worst-case approximation ratios for the heuristics and test their performance by comparing the solutions to the optimal solutions for small instances. In addition, we apply our heuristics to several artificial instances with different characteristics and sizes. Our experiments show that substantial savings are possible with this concept compared to truck-only delivery.

The online appendix is available at https://doi.org/10.1287/trsc.2017.0791 .

/pdf/optimization-approaches-for-the-traveling-salesman-problem-oh3yx9f3yu.pdf

Optimization Approaches for the Traveling Salesman Problem with Drone

The fast and cost-effcient home delivery of goods ordered online is logistically challenging. Many companies are looking for new ways to cross the last-mile to their customers. One technology-enabled opportunity that recently has received much attention is the use of a drone to support deliveries. An innovative last-mile delivery concept in which a truck collaborates with a drone to make deliveries gives rise to a new variant of the traveling salesman problem (TSP) that we call the TSP with drone. In this paper, we formulate this problem as an MIP model and develop several fast route first-cluster second heuristics based on local search and dynamic programming. We prove worst-case approximation ratios for the heuristics and test their performance by comparing the solutions to the optimal solutions for small instances. In addition, we apply our heuristics to several artificial instances with different characteristics and sizes. Our numerical analysis shows that substantial savings are possible with this concept in comparison to truck-only delivery.

/pdf/optimization-approaches-for-the-traveling-salesman-problem-1u5pxh6k9c.pdf

The emergence of the Internet of things and the widespread deployment of diverse computing systems have led to the formation of heterogeneous multi-agent systems (MAS) to complete a variety of tasks. Motivated to highlight the state of the art on existing MAS while identifying their limitations, remaining challenges, and possible future directions, we survey recent contributions to the field. We focus on robot agents and emphasize the challenges of MAS sub-fields including task decomposition, coalition formation, task allocation, perception, and multi-agent planning and control. While some components have seen more advancements than others, more research is required before effective autonomous MAS can be deployed in real smart city settings that are less restrictive than the assumed validation environments of MAS. Specifically, more autonomous end-to-end solutions need to be experimentally tested and developed while incorporating natural language ontology and dictionaries to automate complex task decomposition and leveraging big data advancements to improve perception algorithms for robotics.

Cooperative Heterogeneous Multi-Robot Systems: A Survey

This paper addresses a search planning and task allocation problem for a Unmanned Aerial Vehicle (UAV) team that performs a search and destroy mission in an environment where targets with different values move around. The UAVs are heterogeneous having different sensing and attack capabilities, and carry limited amount of munitions. The objective of the mission is to find targets and eliminate them as quickly as possible considering the values of the targets. In this context, there are two distinct issues that need to be addressed simultaneously: search planning and task allocation. The search plan generates an efficient search path for each UAV to facilitate a fast target detection. The task allocation assigns UAVs attack tasks over detected targets such that each UAV's attack capability is respected. We model these two issues in one framework and propose a distributed approach that utilizes a probabilistic decision making mechanism based on response threshold model. The proposed approach accounts for natural uncertainties in the environment, and provides flexibility, resulting in efficient exploration in the environment and effective allocation of attack tasks. The approach is evaluated in simulation experiments in comparison with other methods, of which results show that our approach outperforms the other methods.

Response Threshold Model Based UAV Search Planning and Task Allocation

This paper addresses a task allocation problem for a team of UAVs that cooperatively performs a search and attack mission in an unknown region. The UAVs are heterogeneous carrying different types and amounts of munition resources, and limited in communications and sensing capabilities. The environment is highly uncertain and dynamic where no prior information is available and dynamic events such as UAV failures unpredictably occur. The objective of the mission is to maximize total reward obtained by destroying targets within a given mission horizon. A group of UAVs may need to be formed to attack a target because individual UAVs may not have sufficient resources for the execution of attack tasks. Instantaneous task allocation approaches that seek for optimal solution for current tasks cannot effectively account for the unpredictability of future tasks in the uncertain dynamic environment. In this paper, we propose a distributed task allocation scheme based on resource welfare of which concept is adopted from economics. The approach we present enables the UAV team to effectively utilize resources by balancing resource depletions and consequently be capable of smoothly responding to dynamic events by retaining more UAVs available. Simulation experiments were conducted in various conditions to evaluate the performance of the proposed approach in comparison with the instantaneous task allocation method. The results show that our approach improves the performance by up to 29.3 % with respect to the instantaneous task allocation method.

Resource Welfare Based Task Allocation for UAV Team with Resource Constraints

To detect faults on the power transmission and distribution systems, current electric utilities perform a visual inspection by dispatching line crews and helicopters. This practice has disadvantages such as high operation costs and safety concerns. To resolve these issues, power utilities are considering the use of an unmanned aircraft system (UAS). In this paper, we formulate an optimization model to find an efficient flight path for a UAS for visually inspecting a transmission tower. The objective of the model is to maximize a function involving three performance ratios, namely, flight time, image quality, and tower coverage. The optimization model is non-linear, non-differentiable, and multi-modal. We solve the problem by using a particle swarm optimization (PSO) based-algorithm and a simulated annealing (SA) based-algorithm and compare their results. We test the model under three inspection strategies. The experimental results show that the PSO-based algorithm outperforms the SA-based algorithm. They also show that the proposed model can provide a flight path that comprises a good balance over the three performance ratios.

Unmanned Aircraft System Path Planning for Visually Inspecting Electric Transmission Towers

The use of wind turbines to generate electricity is growing worldwide. They comprise an extended area of hundreds of square miles, making the inspection process difficult and time-consuming. Recently, there has been an increasing interest in using a drone, or also known as unmanned aircraft systems, for inspecting wind turbines. Motivated by leveraging drone technology, this paper provides a routing optimization model to reduce the total operation time for inspecting a wind farm. We assume that one drone and one ground vehicle which carries the drone and extra batteries and charging equipment are available. The optimization model is solved in two steps. The first step clusters the wind turbines and optimizes the drone routing in each cluster by solving the classical traveling salesman problem using an integer linear programming model. The second step optimizes the ground vehicle routing by solving the equality generalized traveling salesman problem using an integer linear programming model. We test our proposed model using three case studies created by using actual wind farm locations. We compare the results with two models. One model assumes no clustering of the wind turbines, and the other model uses a greedy approach for determining the ground vehicle route. The results show that the proposed model is more efficient at different flight speeds and endurances. Also, we confirm that the efficiency increases as the drone flies faster or it has longer flight endurance.

Hyeoncheol Baik

Papers

Response Threshold Model Based UAV Search Planning and Task Allocation

Resource Welfare Based Task Allocation for UAV Team with Resource Constraints

Unmanned Aircraft System Path Planning for Visually Inspecting Electric Transmission Towers

An optimization drone routing model for inspecting wind farms

Unbalanced data, type II error, and nonlinearity in predicting M&A failure