Naoto Mukai

Optimization of Vehicle Assignment for Car Sharing System

Recent advances in information technology enable realization of new on-demand transportation systems. A car sharing system is one of the new on-demand transportation systems and its concept is that people share vehicles to save maintenance cost. A car sharing system is expected as a solution for traffic jams and lack of parkings. There are two types of car sharing systems: the one-way type and the round-trip type. In this paper, we focus on the one-way type. In the one-way type, users can return vehicles to any stations at any time. Thus, it is hard to keep distribution balance of parked vehicles among stations. We propose a method for optimizing vehicle assignment according to distribution balance of parked vehicles. Finally, we report experimental results of our simulation, and evaluate the effect of our method.

Classification on nonlinear mapping of reducts based on nearest neighbor relation

Dimension reduction of data is an important theme in the data processing and on the web to represent and manipulate higher dimensional data. Rough set is fundamental and useful to process higher dimensional data. Reduct in the rough set is a minimal subset of features, which has the same discernible power as the entire features in the higher dimensional scheme. It is shown that nearest neighbor relation with minimal distance introduced here has a basic information for classification. In this paper, a new reduct generation method based on the nearest neighbor relation with minimal distance is proposed. To improve the classification accuracy of reducts, we develop a nonlinear mapping method on the nearest neighbor relation, which makes vector data relation among neighbor data and preserves data ordering.

Adaptive traffic signal control based on vehicle route sharing by wireless communication

In these years, a problem of traffic congestion is a growing concern due to increasing vehicle holders in urban areas. One of key issues to ease the traffic congestion is a traffic signal control. In this paper, we propose a control system for traffic signals by a vehicle route sharing. The vehicle route sharing is to share position and path information (i.e., probe data) among vehicles. The shared information is used for calculating a criteria of traffic congestion called expected traffic congestion. The parameters of traffic signals (i.e., cycle, split, and offset) are optimized on the basis of the expected traffic congestion. We performed a multiagent simulation to evaluate the effectiveness of our proposed system. The results show that our proposed method can ease traffic congestion effectively compared with other traditional methods.

R-Tree Based Path Representation for Vehicle Routing Problem

Vehicle Routing Problem (VRP) is a problem to find the minimum path length for a fleet of vehicles which deliver goods from the depot according to demands. Some specific methods for the VRP include evolutionary approaches such as Genetic Algorithms. A naive data structure (i.e., chromosome) for the evolutionary approaches is called Path Representation (PR) which is an ordered list structure which represents a single path. On the other hand, we propose a novel data structure called R-Tree based Path Representation (R-PR). R-PR is a tree structure based on a spatial indexing structure called R-Tree. Each node of R-PR represents a sub-path (i.e., a partial solution) for VRP, and its parent node represents a set of the sub-paths, hierarchically. We compare R-PR with PR by using some of VRP instances by Augerat et al, and the results show that R-PR outperforms PR in large scale instances.

Route optimisation using evolutionary approaches for on-demand pickup problem

The development of information technologies realises an on-demand transport (pick-up) system. In this paper, we simulate transport situations for the system based on multi-agent model to find efficient strategies. We examine four types of driver agents; random agent, greedy agent, Q-learning agent, and Genetic agent. Random agent and Greedy agents select the next pick-up points from its surround without learning and optimisation. In contrast, Q-learning agent estimates the expectation value of pick-up quantity by Q-learning, and Genetic agent optimises its travel routes by Genetic algorithm. Finally, we report our experimental results to evaluate the effect of the four strategies.

Simulation evaluation for on-demand bus system with electrical vehicles

In this paper, we focus on a new transport system called on-demand bus system which is introduced on a trial basis to local cities in Japan. In the system, share-ride buses transport customers door-to-door according to users requests. A user can specify the position and time to get the bus in the service area, thus the on-demand bus is more flexible and profitable system compared to traditional transport systems (i.e., fixed route bus). Electrical vehicles are also attracting attention as a new transportation device in these years. The electric vehicles are environmentally friendly because they produce zero emissions and do not pollute the air. However, there is some issues to be solved for practical use of electrical vehicles, i.e., the price of charger and the mileage per charge. Therefore, we adopt an evolutionary approach to solve a path optimization problem for the on-demand bus with electrical vehicles. It is very important to reduce the amount of recharge time for effective operation of electrical vehicles. An evolutionary algorithm minimizes the traveling distance of vehicles by a mutation operation (i.e., the exchange of sub-routes of vehicles) in order to reduce the amount of recharge time. We will show some comparison experiments by computer simulation, and show the performance of our algorithm for the on-demand bus with electrical vehicles.

Simulation analysis of decision-making policy for on-demand transport systems

In recent years, on-demand transport systems (such as a demand-bus system) are focused as a new transport service in Japan. An on-demand vehicle visits pick-up and delivery points by door-to-door according to the occurrences of requests. This service can be regarded as a cooperative (or competitive) profit problem among transport vehicles. Thus, a decision-making for the problem is an important factor for the profits of vehicles (i.e., drivers). However, it is difficult to find an optimal solution of the problem, because there are some uncertain risks, e.g., the occurrence probability of requests and the selfishness of other rival vehicles. Therefore, this paper proposes a transport policy for on-demand vehicles to control the uncertain risks. First, we classify the profit of vehicles as “assured profit” and “potential profit”. Second, we propose a “profit policy” and “selection policy” based on the classification of the profits. Moreover, the selection policy can be classified into “greed”, “mixed”, “competitive”, and “cooperative”. These selection policies are represented by selection probabilities of the next visit points to cooperate or compete with other vehicles. Finally, we report simulation results and analyze the effectiveness of our proposal policies.

Nonlinear Mapping of Reducts - Nearest Neighbor Classification

Dimension reduction of data is an important theme in the data processing. Reduct in the rough set is useful which has the same discernible power as the entire features in the higher dimensional scheme. But, classification with higher accuracy is not obtained in the reduct followed by nearest neighbor processing. To attack the problem, it is shown that nearest neighbor relation with minimal distance introduced here has a basic information for classification. In this paper, a new reduct generation method based on the nearest neighbor relation with minimal distance is proposed. To improve the classification accuracy of reducts, we develop a nonlinear mapping method on the nearest neighbor relation, which makes vector data relation among neighbor data and preserves data ordering.

Route Optimization Using Q-Learning for On-Demand Bus Systems

In this paper, we focus on a new transport service called on-demand bus system. A major feature of the system is that buses pick up customers door-to-door when needed or required. Thus, there is no pre-determined travel routes for buses, and travel routes must be changed according to the occurrence frequency of customers. In order to find a more effective travel plan to the problem, we adopt Q-learning which is one of the machine learning algorithms. However, native Q-learning is inadequate to our target problem because the number of customers at pick-up points is time-dependent. Therefore, we improve an update process of Q values and a selection process of the next pick-up point, on the basis of time passage parameters. In particular, rewards are understated in update process, on the other hand, Q values are overstated in selection process. At the last, we report our simulation results and show the effectiveness of our algorithm for the problem.