Mobile robot

SLAM (Simultaneous Localization and Mapping) system with Four RGB-D cameras

A SLAM (Simultaneous Localization and Mapping) system utilizing four RGB-D cameras (Microsoft Kinect) has been developed to capture 3D environmental structures in real time as the robot moves. For accurate localization, the system employs Collaborative Probabilistic SLAM (CPS-SLAM), in which the robot's position is estimated by a parent robot using a total station (laser range finder).

SLAM using four kinect cameras photo System configuration photo Measurement results

Papers

Fast 3D localization for mobile robot using Normal Distributions Transform

We propose an efficient 3D global localization and tracking method for mobile robots operating in large-scale environments, using 3D geometric maps and RGB-D cameras. With the rapid advancement of high-resolution 3D range sensors, the need for high-speed processing of large-scale 3D data has become a critical challenge in robotic applications such as localization. To address this issue, the proposed method employs a Normal Distributions (ND) voxel representation. First, a 3D geometric map represented by point clouds is converted into multiple ND voxels, from which local features are extracted and stored as an environmental map. Similarly, range data captured by an RGB-D camera is also transformed into ND voxels, and corresponding local features are computed. For global localization and tracking, the similarity between ND voxels from the environmental map and those from the sensory data is evaluated using either local feature matching or the Kullback–Leibler divergence. The optimal robot pose is then estimated within a particle filter framework.

Kinect color image Kinect depth image Localization process

Papers

Quadcopter Helicopter

We are developing quadcopters and helicopters for use in land surveying applications.

Helicopter Helicopter Helicopter
photo Quadcopter photo Quadcopter Laser scanning from quadcopter

Papers

Spatial change detection using voxel classification by normal distributions transform

Detecting spatial changes in the environment surrounding a robot is critical for various robotic applications, including search and rescue, security, and surveillance. This paper presents a fast spatial change detection method for mobile robots, which utilizes an on-board RGB-D or stereo camera in conjunction with a high-precision 3D map generated by a laser scanner. In the proposed method, both the map and real-time sensor data are converted into grid-based representations—Normal Distributions (ND) voxels—using normal distribution transformation. These ND voxels are then classified into three categories based on their characteristics. The map and sensor data are compared using these categories and the statistical features of the ND voxels. To further enhance robustness, overlapping and voting techniques are introduced. The proposed real-time localization and spatial change detection methods were validated through experiments conducted in both indoor and outdoor environments using a mobile robot equipped with real-time range sensors.

photo Mobile robot with Kinect Detected differences ICRA 2019 Video

Papers

Tour guide robot and personal mobility vehicle

We are developing a tour guide robot and a personal mobility vehicle that utilize not only their onboard sensors but also external sensors installed along streets.

Personal vehicle Wheelchair-type personal vehicle
photo
Tour guide robot, Qurin
photo
Tour guide robot, Qurin 2
Tuor guide robot Navigation using QZSS and 5G
Real time transmission of 4K 360 degrees video via 5G network Navigation using LiDAR and 5G
Tour guide experiment in theme park Guide experiment in hospital

Papers

Autonomous lawn-mowing robot

We are developing an autonomous lawn-mowing robot equipped with QZSS MICHIBIKI for centimeter-level positioning via CLAS (Centimeter Level Augmentation Service), and a 3D LiDAR sensor for obstacle detection. For autonomous path planning and motion control toward multiple targets, we employ the Nav2 framework in ROS2.

photo
1st model autonomous lawn-mowing robot
photo
2nd model autonomous lawn-mowing robot

Papers

Automatic illuminance measurement multiple robots

We are developing an automatic illuminance measurement multi-robot system designed to assess lighting conditions in large-scale warehouses. This system employs advanced sensors and algorithms to ensure optimal illuminance levels throughout expansive areas, thereby enhancing both energy efficiency and operational safety.

photo
Measurement by automatic illuminance measurement multiple robots
photo
Automatic illuminance measurement multiple robots
Illuminance measurement Illuminance measurement

Papers

Beach cleaning robot

We are developing a beach-cleaning robot designed to collect microplastic debris. The goal is to address the widespread problem of microplastic pollution in marine environments and to help ensure cleaner, safer beaches for both humans and wildlife.

photo
Beach cleaning robot
photo
Beach measurement robot
Beach cleaning Beach cleaning

Papers

  • 海洋破砕プラスチックごみ回収ロボットの開発
    宇野 光輝, 倉爪 亮
    第23回計測自動制御学会システムインテグレーション部門講演会 SI2022, 1A2-D04, 2022.12.14-16 [pdf][bibtex]
  • 海洋破砕プラスチックごみ回収ロボットシステムの開発 -レーザースキャナの反射輝度によるごみ検出とロボットの誘導-
    有瀬 昌矢, 松本 耕平, 倉爪 亮
    第23回計測自動制御学会システムインテグレーション部門講演会 SI2022, 1A2-D11, 2022.12.14-16 [pdf][bibtex]
  • 海洋破砕プラスチックごみ回収機構の開発
    宇野 光輝, 倉爪 亮
    第22回計測自動制御学会システムインテグレーション部門講演会 SI2021, 1H4-03, 2021.12.15-17 [pdf][bibtex]
  • 海洋破砕プラスチックごみ回収ロボットシステムの開発 レーザスキャナの反射輝度を用いた海岸環境の識別
    有瀬 昌矢, 倉爪 亮
    日本機械学会ロボティクスメカトロニクス講演会2021, 1P2-G09, 2021.6.6-8
    [pdf][bibtex]

Construction robot

We are developing a retrofit-type remotely controlled backhoe, a compaction evaluation method for vibratory rollers using the multi-sensor terminal “Sensor Pod,” and a cyber-physical platform for construction sites called “ROS2-TMS for Construction.”

A retrofit-type remotely controlled backhoe photo
Compaction evaluation experiment by vibratory rollers using Sensor Pods

Papers

  • Evaluation of ground stiffness using multiple accelerometers on the ground during compaction by vibratory rollers
    Yusuke Tamaishi, Kentaro Fukuda, Kazuto Nakashima, Ryuichi Maeda, Kohei Matsumoto, Ryo Kurazume
    40th International Symposium on Automation and Robotics in Construction (ISARC 2023), pp. ,doi:, Chennai, 2023.7.4-7, 2023
    [pdf][bibtex]
  • 土木工事における地盤剛性評価・安全管理のための分散型センサポッドの開発
    福田 健太郎, 中嶋 一斗, 前田 龍一, 松本 耕平, 倉爪 亮
    第23回ロボティクスシンポジア, , 2023.3.15-16 [pdf][bibtex]
  • レトロフィット型バックホウ遠隔操縦システムの開発 -第2報 操作性の向上と後付センシングシステムの開発-
    柴田 航志, 西浦 悠生, 倉爪 亮
    第23回計測自動制御学会システムインテグレーション部門講演会 SI2022, 3P2-G17, 2022.12.14-16 [pdf][bibtex]
  • 多点同期振動データの波形歪みに基づく地盤剛性評価手法の提案
    福田 健太郎, 中嶋 一斗, 倉爪 亮
    第20回 建設ロボットシンポジウム, P2-2, 2022.8.24-25
    [pdf][bibtex]
  • 多様な環境に適応しインフラ構築を革新する協働AIロボット(土工を革新するハードウエアと現場を俯瞰するセンサポッドシステムの紹介)
    永谷圭司, 大須賀公一, 竹囲年延, 倉爪亮
    第4回 i-Constructionの推進に関するシンポジウム, pp.115-118, 2022.7.11
    [pdf][bibtex]
  • レトロフィット型バックホウ遠隔操縦システムの開発
    西浦 悠生, 中嶋 一斗, 倉爪 亮
    日本機械学会ロボティクスメカトロニクス講演会2022, 1P1-C07, 2022.6.1-4
    [pdf][bibtex]
  • 転圧地盤評価のための分散型センサポッドの開発 -第2 報多点同期振動データの波形歪みに基づく地盤剛性の定量化-
    福田健太郎, 中嶋 一斗, 倉爪 亮
    日本機械学会ロボティクスメカトロニクス講演会2022, 1A1-E04, 2022.6.1-4
    [pdf][bibtex]
  • 転圧地盤評価のための分散型センサポッドの開発
    福田 健太郎, 中嶋 一斗, 倉爪 亮
    第22回計測自動制御学会システムインテグレーション部門講演会 SI2021, 3H4-05, 2021.12.15-17 [pdf][bibtex]