Absolute Pose Estimation of a Stair Cleaning Robot using Staircase Geometry and ToF sensors Robotics & AI (RoAI) 2019-12-25 | sTetro (stair Tetris) is a stair cleaning robot which can climb the staircase autonomously with its shape-shifting capabilities. As this robot is intended to traverse multi-story buildings, hence its positioning/ localization information is an essential component of the overall system. Usually, the indoor mobile robots rely on some external system for localization information, e.g. WiFi, UWB, vision, RFID signals, or other forms of indoor pseudolites (GPS-like) system. This requires the installation of additional hardware and/or modification of the working environment for precise positioning information of a mobile platform. As the dimensions of the staircase are usually known a priori, this information can be used to localize the sTetro robot on the stairs. In this article, the geometry information of the staircase has been exploited to estimate the absolute position of the robot in 3D space with measurements from the onboard time-of-flight (ToF) range sensors only. The heading angle of the robot is also estimated with measurements from two ToF sensors installed in front of the sTetro robot. Results achieved by conducting experiments on different types of the staircase with real robot prove the efficacy of the proposed technique.
Autonomous Multi-Floor Navigation by a Quadruped Robot in Infrastructure-Free Buildings via elevator Robotics & AI (RoAI) 2025-06-17 | This project presents a fully autonomous elevator interaction frameworkfor quadruped robots operating in un-instrumented, multi-floorbuildings—without requiring IoT connectivity, infrastructuremodifications, or human intervention. Unlike prior work onwheeled platforms or API-dependent systems, our approachenables a legged robot to perceive, approach, and physicallyinteract with standard elevator panels using only onboard sensingand manipulation. The system integrates YOLOv11-based paneldetection, AprilTag-guided spatial alignment, and a 4-DOF armfor precise button pressing within a modular ROS2 architecture.Our robot autonomously completes all stages of elevator use:external panel detection, entry, destination selection, and post-exitnavigation. Real-world trials in a four-story public buildingyielded consistent docking, sub-cm level button detection and press accuracy, and 100 percent verified activation across multiple runs, setting a newbenchmark for practical, infrastructure-free vertical mobility inrobotics.
Introduction to Mobile Robots-02 Robotics & AI (RoAI) 2025-03-12 | Types of MR. Holonomic and Non-holonomic MR. MR Motions
Path Planning-Part 2 Robotics & AI (RoAI) 2025-03-12 | Path planning on grid maps. Grassfire algorithm examples of grid map
Introduction to Path Planning in Mobile Robots-Part 1 Robotics & AI (RoAI) 2025-03-12 | Introduction path planning algorithms. Graph traversal algorithms. Grassfire algorithms. Occupancy grid map. Examples of grassfire algo on grid map
Introduction to Mobile Robots-01 Robotics & AI (RoAI) 2025-03-12 | Introduction to mobile robots. Comparison with manipulators. Types of MR.
PART 02: Forward and Inv. Differential Kin. Models of 2DoF Robotic Manipulator Robotics & AI (RoAI) 2025-02-24 | This is PART-2 of class lec. of "Introduction to Robotics - Spring 2025", at School of IT & Engineering (SiTE), Kazakh-British Technical University (KBTU). In this lecture i will explain, step-by-step, all the process of deriving Differential Kinematics models of 2DoF robotic manipulator, with a concrete example
PART 01 Forward Inv._Kinematic Model_Forward Inv._Diff. Kin. Models of 2DoF Robotic Manipulator Robotics & AI (RoAI) 2025-02-24 | This is actual class lec. of "Introduction to Robotics - Spring 2025", at School of IT & Engineering (SiTE), Kazakh-British Technical University (KBTU). In this lecture i will explain, step-by-step, all the process of deriving Kinematics models of 2DoF robotic manipulator, with a concrete example.
BIM-Based Navigation Robotics & AI (RoAI) 2023-06-02 | In this work, we present a technique to leverage these BIM for robotic navigation, instead of the typical SLAM map (e.g. gmapping, hectorSlam etc), created by scanning the environment beforehand by using mobile robot. The proposed system describes a BIM-based robot navigation system designed to operate in indoor environments. The system enables the user to supply the robot with goal position(s) and the robot navigates to the desired locations by avoiding obstacles, thus eliminating the need for pre-exploration/mapping of the environment, which is a big advantage for many inaccessible environments.
Controller Design MS Course at KBTU Robotics & AI (RoAI) 2023-05-23 | Discuss and Design different controllers and Compensators in the course of "Feedback Control system analysis & Design" at KBTU.
Human_Follower_using ArUco marker Detection Robotics & AI (RoAI) 2021-10-07 | I used a simple yet effective technique for Human Following. An aruco marker is a fiducial marker that is placed on the object being imaged. Distance and angle of Line/vector joining two frame's origin (camera frame and arUco marker frame) is calculated and passed to ROS onboard to control Robot motion.
BIM-based Multi_goal_point demo Robotics & AI (RoAI) 2021-01-17 | The robot takes BIM information and Generates Goal Points for autonomous navigation, to inspect/Detect particular building component(s) in the working environment.
Initialization Position in BIM env. using Range, Bearing from Lidar Robotics & AI (RoAI) 2020-08-12 | ...
Multi-Goal Point Navigation Robotics & AI (RoAI) 2020-08-07 | The user can select/choose multiple objects(goal points:GP) in Building information model(BIM ) for inspection (according to inspection List)The robot will visit each goal point one-by-one & give live video feedback.Object Detectors (e.g. YOLO, R-CNN) can be applied to monitor the building elements & generate Status of each GP automatically.This way robot visits specified GPs with minimal human intervention.
2D Floor Plan Navigation Robotics & AI (RoAI) 2020-07-17 | Goal: To obtain 2D Map directly from BIM modelOpen 3D BIM model( “model.rvt” ) in Autodesk Revit AppDo some editing (e.g. remove extra things, like furniture etc).See the Top View, and save it as 2D floor plane as “floor.png” image file.Now, open Python code to Convert an image foor plane to ROS map (as I did for CAD-based map).The difference between CAD map & 2D floor map is that in to obtain CAD map, we manually draw 2d floor plan in Autocad, however, here we can export 2D floor plan directly from BIM.
ROS BIM Robotics & AI (RoAI) 2020-07-09 | Goal: To obtain 3D Octo_Map from BIM modelOpen 3D BIM model( “model.rvt” ) in Autodesk Revit App. Do some editing (e.g. remove extra things, like furniture etc). Save this model as “model.dwg” file. Open this model in SketchUp App. To add/remove material type (if required) and save it as “model.dae” .Now import this model.dae in Gazebo.Run simulated robot in Gazebo world & generate 3D map of the 3D BIM model.
OctoMapGenerated Robotics & AI (RoAI) 2020-06-26 | OctoMap is an Efficient Probabilistic 3D Mapping Framework Based on Octrees[1].Full 3D model of the Env: The map generated is able to model arbitrary environments without prior assumptions about it. The representation models occupied areas as well as free space. Unknown areas of the environment are implicitly encoded in the map. While the distinction between free space and occupied space is essential for safe robot navigation, information about unknown areas is important, e.g., for autonomous exploration of an environment.Methodology:I installed the required Packages for 3D map, e.g. octomap_mapping [2], octomap_rviz_plugins [3], turtlebot3_teleop[4] etc.I imported Gazebo model of Robotics Lab I (B6- lab), whose 3D is to be built.Ran robot in the evn. manually by controlling the robot from Keyboard.Built the 3D Occupancy grid map (Octomap) and saved it for future use in navigation and path planning (will show next week)Ref. 1. Hornung, Armin, et al. "OctoMap: An efficient probabilistic 3D mapping framework based on octrees." Autonomous robots 34.3 (2013): 189-206.2. http://wiki.ros.org/octomap3. http://wiki.ros.org/octomap_rviz_plugins4. http://wiki.ros.org/turtlebot3_teleop5. ark.intel.com/content/www/us/en/ark/products/92256/intel-realsense-camera-r200.html6. emanual.robotis.com/docs/en/platform/turtlebot3/specifications
Understanding BIM Autodesk Revit Robotics & AI (RoAI) 2020-06-12 | We can open BIM models in Autodesk Revit and view and analysis building element properties. Further, for using BIM model for ground mobile robot nav,. we can export floor-planes from BIM models.
Testing LOC Algo Robotics & AI (RoAI) 2020-05-08 | Initially the Robot in simulated env is at unknown position.It appears in rviz at a random position.Robot detects an Unique feature (cylinder in this case, at Known Position, and Orientation in BIM frame) using its Lidar (i.e. robot gets range and angle measurements to the feature).Robot calculates its position(x,y) using LOC algorithm and sends it to rviz.rviz gets initialisation data and places robot’s visualization icon at that initial position.Then we can start autonomous nav as usual.
Intro Lecture Demo Robotics & AI (RoAI) 2020-05-08 | Self Introduction and Demo Lecture by Dr Muhammad Ilyas
Init pos Calculation in ROS Robotics & AI (RoAI) 2020-04-30 | Position & Orientation (pose) Estimation Algorithm using BIM info. & Lidar measurements
init pose calculation Robotics & AI (RoAI) 2020-04-30 | The robot detects a unique feature in its working env and calculates its own pose , using known global pose of that feature.
Sending Goal Pose Programetically Robotics & AI (RoAI) 2020-04-24 | This videos shows an automated Goal (using python code) position & Orientation sent to a mobile robot (instead of using rviz's buttons). I created a Node that uses "SimpleActionClient" server and sends goal pose to move_base. The robot gets goal pose data and planes path to reach the goal position autonomously.
send goal programetically Robotics & AI (RoAI) 2020-04-22 | This videos shows an automated Goal (using python code) position & Orientation sent to a mobile robot (instead of using rviz's buttons). I created a Node that uses "SimpleActionClient" server and sends goal pose to move_base. The robot gets goal pose data and planes path to reach the goal position autonomously.
send goal Pose Programetically Robotics & AI (RoAI) 2020-04-22 | This videos shows an automated Goal pose(position & Orientation) sent to a mobile robot (instead using rviz's buttons). I created a Node that uses "SimpleActionClient" server and sends goal pose to move_base. The robot gets goal pose data and planes path to reach goal position autonomously.
Programatically Setting of Robots Initial Pose Robotics & AI (RoAI) 2020-04-17 | Traditionally, after launching rviz in ROS, the initial position of robot is set using “2D Pose Estimate” button in rviz manually. It causes errors in setting initial position, which is critical for further path planning/navigation. I have automated this process of Pose_Initialization programmatically, which sets exactly the same initial position as robot’s current Pose in real world.
CAD based navigation demo Robotics & AI (RoAI) 2020-03-04 | First attempt to Nav a physical robot based on CAD drawing.
CAD-drawing based Mobile robot navigation Robotics & AI (RoAI) 2019-11-28 | CAD-Based Indoor Mobile Robot Navigation :In traditional indoor mobile robot navigation, Map is generated using gmapping SLAM/scanning process. In many cases, creating maps pre-handed is not possible/feasible. However, CAD drawings (or BIM models) are available for most of modern buildings now-a-days.In our approach, we use CAD drawings for indoor mobile robot’s autonomous navigation.It precludes the necessity of creating Maps of the environments using scanning methods, or gmapping SLAM, which is a tedious and costly process.(in this video, we used CAD dwng of our lab (control systems lab) in NTU & managed to navigate simulated mobile robot around).Contact: milyasmeo@gmail.com ( IDr LYAS MUHAMMAD, Research Fellow NTU, Singapore )
CAD Based Navigation Robotics & AI (RoAI) 2019-11-11 | In traditional indoor mobile robot navigation, Map is generated using gmapping SLAM/scanning process In many cases, creating maps pre-handed is not possible/feasible. However, CAD drawings are available for most of modern buildings now a days.In our approach, we use CAD drawings for indoor mobile robot’s autonomous navigation.It precludes the necessity of creating Maps of the environments using scanning methods, or gmapping SLAM, which is tedious and costly methods.
CAD Based Navigation Robotics & AI (RoAI) 2019-11-09 | CAD-Based Indoor Mobile Robot Navigation :In traditional indoor mobile robot navigation, Map is generated using gmapping SLAM/scanning process. In many cases, creating maps pre-handed is not possible/feasible. However, CAD drawings (or BIM models) are available for most of modern buildings now-a-days.In our approach, we use CAD drawings for indoor mobile robot’s autonomous navigation.It precludes the necessity of creating Maps of the environments using scanning methods, or gmapping SLAM, which is a tedious and costly process.Contact: milyasmeo@gmail.com ( ILYAS MUHAMMAD )
