“Each lighting fixture has its own characteristic flickering pattern, which is distinguishable from others” — This is the fact the authors had revealed in their previous works. Then, there is a scheme of indoor self-localization by using it. The author call it “CEPHEID”, which is an acronym for “Ceiling Embedded PHoto Echo ID”. In this paper, the author introduces CEPHEID and describes its details. In the method, each lighting fixture is identified by its own characteristic vector, that is generated by its flickering pattern in range of acoustic frequency band. Then, the characteristic vectors are classified by 1D CNN (Convolutional Neural Network). The validity of the idea is shown by experiment results.
In the Rescue Robot Contest (ResCon), rescue dummies, which is called as ‘Damiyan’, are employed for competition evaluation. Rescue teams are required to rescue Damiyan gently and quickly. In addition, it is also required to identify the signs that are presented, and these are the evaluation items. Therefore, Damiyan needs various sensors to quantify “neck bending”, “pressure on body”, “posture” and “impact” as evaluation indicators of gentleness. In addition, the functions of “presenting color”, “uttering sound” and “presenting two-dimensional code” are implemented as the presentation of signs to be identified. To achieve this, we have developed a new Damiyan that incorporates the microprocessor BlueNinja as a core system for sensing, communication, and overall system control. In addition, an airtight body was manufactured to measure the grip strength, which measures the pressure on the body. The body is molded using a 3D printer and flexible filaments, and the surface is coated with a solvent to ensure airtightness. By using this new Damiyan, we achieved stable operation and reduction of management operations. In this report, details of the concept, the policy and the process of developing this newer Damiyan are described.
In this study, the variable friction dumper driven with pneumatic actuator is developed to correct body movement. The developed device is constructed with friction members which are made with rubber, and pneumatic artificial rubber muscle, nylon belt. By controlling inner pressure of rubber muscle or changing number of rubber muscles, nylon belt pull-out force generated from friction force between friction members can be changed. It is assumed that this pull-out force is used to correct movement. In this paper, structure and principle of this device are described and then the characteristics are discussed.
Industrial robots are required to recover from errors autonomously under uncertain environment. In this paper, we propose a recovery action planning system by considering semantic information behind the detected error information. The proposed system uses Conceptual Graph to classify errors and Bayesian Network to evaluate the uncertainty in oeder to determine feasible repair strategies. We demonstrate the effectiveness of the proposed decision model by simulations of an assembly task in which actions of multiple robots affect each other.
We developed a new method for obstacles detection and 3D reconstruction using a 3D map. Obstacles detection and 3D reconstruction are key functions of autonomous driving. It is easy to detect and reconstruct static obstacles three-dimensionally because they exist in the 3D map. However, the detection and the 3D reconstruction of dynamic obstacles that are not in the 3D map is difficult for a typical in-vehicle camera that cannot measure the distance. We aim to detect dynamic obstacles three-dimensionally, using an in-vehicle camera. And we deal with the new problem of accurate 3D reconstruction by using a monocular camera and a 3D map. To solve this problem, we focused on semantic segmentation for detection and depth completion to complement the depth map. We propose a multi-task neural network (NN) that shares the encoder of semantic segmentation NN and depth completion NN, whose inputs are an image and the 3D map. The proposed multi-task NN detects dynamic obstacles 1.4 times more accurately than the single-task state-of-the-art method.
One of the tasks in operating a robot autonomously is to identify road conditions. For example, when the robot navigates outdoors, such as in a park, it is preferable to run on a flat, paved road while avoiding lawns and puddles. In this research, we propose a method to recognize the road conditions using the reflection intensity obtained from the laser range sensor. In the proposed method, the road surface on which the reflection intensity is measured is divided into polar grids, and the distribution of the reflection intensity in each grid is expressed using kernel density estimation. By comparing this distribution with reference data, it is shown that road surface conditions such as lawns, puddles, and paved roads can be identified. In addition, we report the results of experiments using this method in a real environment with existing methods, and report future issues.
In Japan, the number of elderly households facing difficulties in snow removal work is increasing due to the rapid aging and depopulation. Moreover, because of the declining birthrate, human resources needed for the work are insufficient. Therefore the elderly are forced to do the snow removal work on their own, and as a consequence serious accidents may occur during the work. Currently, as a means for solving this problem, there are snow removal substitution services and construction of electric road heating. However, they are not definitive solutions since they are very expensive. The purpose of this study is to develop an autonomous robot system which can perform snow removing with low cost and simple construction. Through the experiments using the prototype, and the cost estimate that assumed practical use, the effectiveness of this system was confirmed.