Masashi Yokozuka

A framework of fault detection algorithm for intelligent vehicle by using real experimental data

Automated vehicles have a possibility to contribute more capacity, safety, low emission and high efficiency to transportation. However, unstable conditions of the automated system can cause serious problem, thus the automated vehicle requires high reliability. The objective of this research is to develop algorithms of fault (unstable condition) detection for automated vehicles, and to improve the overall reliability of the system. In this study, we initially solved and updated identification of some pattern of data constellations under normal and unstable conditions through the experiments in a real world. The multiple experiments were done in the public area (course distance is about 1.1[km]) with some times, where some pedestrian, bicycles and other robots coexisted. The method of detecting faults utilizes mahalanobis distance, correlation coefficient and linearization in order to improve the reliability of detecting the faults correctly, because real experimental conditions include some random noises, and the method must be robust for the changing conditions. The feature of this study is to utilize the experiment results in real world, construct the algorithms and evaluate it. The simulations were done with the real experimental data, in order to evaluate it. The simulation result shows that the proposed system detects faults correctly, and it proves the validity of the proposed method proved.

Sub-Map Dividing and Realignment FastSLAM by Blocking Gibbs MCEM for Large-Scale 3-D Grid Mapping

Abstract In this article, we propose a novel Simultaneous Localization and Mapping (SLAM) method by using a sampling-based approach. FastSLAM is well-known approach as a sampling-based SLAM method. FastSLAM utilizes a theorem that map errors are decidable under a sample of trajectories. From this theorem, FastSLAM samples many trajectories and maps to find a minimum error map. However, in case of constructing a large-scale grid map, FastSLAM becomes unuseful since the method requires huge memory to generate many grid maps. Our proposed method requires only one map that is deformable corresponding to a trajectory. In order to find a minimum error map, our method only generates trajectories. Our method enables construction of a minimum error map by little memory in comparison with original FastSLAM. Experimental results demonstrate our method is able to construct a large-scale 3-D grid map by low memory usage in comparison with original FastSLAM.

An experimental study on changes of muscle fatigue among traveling by standing-type mobile vehicles and walking

Many robotics technologies are being developed. One of the technologies is utilized in a personal vehicle. A personal vehicle, which has high energy efficiency, has been expected as new personal mobility. It is well-known that a personal vehicle is higher energy efficiency, and requires less space than a normal vehicle. On the other hand, with respect to the fatigue, there are less enough experimental data using a personal vehicle, especially a standing-type personal vehicle. Thus, it will be necessary to evaluate the fatigue using standing-type personal vehicle. Two types of standing-type personal vehicles were used as personal vehicles in this study. The changes of muscle hardness on legs and subjects feeling were employed in order to measure the amount of fatigue. The amount of fatigue was measured before and after moving about 3 km distance in real world on using each vehicle. The experiments and questionnaires after doing experiments were done with 6 subjects. Experimental result shows that there is a difference between two kinds of standing-type personal vehicles. The amount of muscle fatigue by using one small vehicle changes larger than other two methods. It assumed that muscle fatigue can be affected by stability, power and space around legs from the results of experiments and questionnaires.

Accurate depth-map refinement by per-pixel plane fitting for stereo vision

This paper discusses the refinement of sparse and noisy depth-maps to improve stereo measurements. Our method functions as a post-filter for stereo measurements, to remove outliers and interpolate the depths of invalid pixels. Per-pixel plane fitting is employed to estimate the normals of an objects surface in a depth-map. These normals provide information regarding the interpolation of depth and the removal of outliers by evaluating the directions of surfaces. In our experiments, our method successfully reconstructed a dense and accurate geometry from a sparse and noisy depth-map, even where several dozen percent of pixels were outliers and only a few percent were from the original correct geometry. This result indicates a novel method of fast stereo measurement, because dense reconstruction can be performed without stereo matching for all pixels.

Low-cost 3D mobile mapping system by 6 DOF localization using smartphone embedded sensors

To realize auto-driving vehicles transporting to anywhere in a country, a map containing the whole regions is necessary. In order to generate and update the big size map, number of mapping systems is an important matter. To introduce a large number of mapping systems, the costs of a system must be small. This paper describes a low-cost mobile mapping system employing a smartphone for positioning. Although the positioning function of smartphones is accurate enough to detect a self-position for pedestrian navigation, however, the accuracy is not enough to generate maps that contain obstacles for auto-driving. To gain accuracy of the positioning employing a smartphone, this paper uses a bundle adjustment that improves positioning results by fusing multiple sensor data. From experimental results, the proposed low-cost mapping system can generate an accurate map for auto-driving use.

Development of Autonomous Wheelchair for Indoor and Outdoor Traveling

In order to assist elderly people and disabled people, this paper describes development of autonomous wheelchair to travel in indoor and outdoor environments for providing traveling ability to any where. For this aim, the autonomous wheelchairs should have traveling-capability without choosing indoor and outdoor environments. Position detection is a key technology for autonomous driving since people decides a traveling direction from a current position and a destination. GPS is a fundamental technology for position detection. However GPS is not available in indoor cases, and GPS is not always available in outdoor cases when tall buildings occlude satellites. In these cases autonomous wheelchair has to detect a self-position by other sensor systems. In this study we have adopted a localization system utilizing 3D maps and a 3D laser range finder. By the 3D localization system our wheelchair system can detect a self-position robustly if the wheelchair is surrounded by obstacles such as pedestrians. To avoid collision our wheelchair system uses short and long term planning. The short planning finds a safe motion-pattern from every conceivable pattern by the simulation on a map. The long term planning generates a feasible route to destination. If the route generated by the long-term planner collides to some obstacles our wheelchair avoids collision by the short term planning. By the localization system and the planning system our wheelchair could operate in public spaces.

Registration of Low Cost Maps within Large Scale MMS Maps

High resolution 3D mapping of road systems is currently being carried out by expensive Mobile Mapping Systems (MMS) but coverage is limited. Recently Low Cost Sensor (LCS) systems have been developed which use common, low cost, internal MEMS position sensors from mobile phones, but such sensors come with a reduced absolute and relative positional accuracy. This study investigates the registration of LCS maps within MMS maps to improve map coverage and lower costs. MMS and LCS maps of a real world environment are made and registration is performed using feature matching and Iterative Closest Point alignment. Accuracy of ICP alignment is approximately (10cm) and local convergence is possible up to (1m). A combination of feature matching and ICP is used to demonstrate accurate alignment from an initial error of (10m). An example of a LCS map aligned within a MMS map is presented to confirm the use of LCS systems to extend 3D mapping coverage.

Experimental examination and simulation analysis of standing-type personal mobility device sharing

This study considers sharing of standing-type personal mobility devices. Such devices offer various advantages as a means of transportation, but they have been used only slightly for this purpose to date. Their cost makes sharing such devices more attractive than owning them. Sharing personal mobility devices is expected to introduce beneficial social influences such as a modal shift. For this study, we conducted small-scale experiments to assess their feasibility. Then we considered larger cases in simulations. The experiments employ four sharing stations and four personal mobility devices by more than 60 registered users. Larger scale simulations are based on a multi-agent model, presenting results for its usage compared with other means of transportation. Such a simulation is expected to provide a basis for future demand prediction and planning of personal mobility sharing.

Effects of Station Location and Capacity for Personal Mobility Sharing

In this study, we consider the sharing of standing-type personal mobility devices. Such devices have various advantages as a means of transportation, but their cost makes sharing more attractive than owning. Sharing personal mobility devices is expected to have social influences such as a modal shift, however, this is not yet apparent. Consideration of the effects based on a multi-agent simulation is useful, and in a previous study, we showed the preliminary results of such simulation. In this study, we extend the simulation and investigate the effects of the location and capacity of personal mobility sharing stations. The results suggest that a location change largely affects the traveling distance. Such a simulation will provide a basis for the future demand prediction and planning of personal mobility sharing.