We just tested YOLOE (text + visual prompts) from @Ultralytics inside the LumiMerse sim→data pipeline. Early takeaways: - Zero-shot prompts work well on unseen scenes - Real-time inference + segmentation masks - Easy lift into our Unreal/Blender synthetic data loops Why this matters: faster “describe → detect → label” cycles for robotics perception and Sim2Real. #YOLOE #Ultralytics #ComputerVision #Robotics #SyntheticData #UnrealEngine #Blender #Sim2Real #LumiMerse

🚀 Line Following Robot in ROS2 & Gazebo 🤖 Excited to share my latest project — a line-following robot simulation using ROS2 (Humble), Gazebo, and OpenCV! ✅ Detects and follows lines autonomously ✅ Smooth motion with PID controller ✅ Auto-search if the line is lost ✅ Works in simulation or with a real robot Check out the demo and code here: https://lnkd.in/gFzh6vzF #ROS2 #Gazebo #OpenCV #Robotics #AutonomousRobots #PIDController

Autonomous Mapping Demo — ROS 2 + Explore Lite : I’ve successfully implemented autonomous mapping using ROS 2 and the explore_lite package! After initially performing manual mapping with teleop_twist_keyboard, I moved to a fully autonomous setup where the robot explores its environment without any manual control. How it works : SLAM Toolbox builds the map in real time. Explore Lite detects frontiers — edges between known and unknown regions. It automatically sends navigation goals to these unexplored areas. The robot keeps exploring until all reachable frontiers are mapped! In the demo video below, you can see the robot autonomously navigating through the environment while the map expands in RViz — a clear sign of successful frontier-based exploration. Next steps: integrating advanced path planning and obstacle avoidance to improve navigation efficiency. #ROS2 #Robotics #SLAM #ExploreLite #AutonomousMapping #FrontierExploration #Navigation #RoboticsEngineering KIET ROBOTICS LAB ROBOTICS LAB @KIET

The Future Learns to Walk (Cinematic + Thought-Provoking) It doesn’t crawl, it learns. Meet the GHOST MINITAUR — a direct-drive quadruped built by Ghost Robotics. Watch closely: every step is precision, balance, and raw intent. 💡 This isn’t just a robot walking. It’s the moment the future found its footing. Because evolution isn’t over — it’s just gone electric. #Robotics #Innovation #AI #Engineering #FutureTech

I successfully simulated a parallel SCARA robot in #ROS_GAZEBO. Using the rosbridge_server tool, I managed to synchronize the real robot with its simulated counterpart, so that when angles are provided to the real robot, the simulated robot moves in real-time as well. Additionally, in this video, I manually guided the robot along paths considering the presence of an obstacle. These angles are stored via potentiometers and then applied to the robot for repetition. To reduce noise from the potentiometers and account for nonlinear relationships in this robot, I use an #Extended_Kalman_Filter. Finally, the graph of applied angles and the actual executed angles on the robot, as well as the graph of the robot's end-effector performance, can be observed. For controlling and minimizing errors between the input and output angle values, a #PID_controller is employed. #ROS #Gazebo #Robotics

The future of personal mobility might just have four legs. This incredible video shows a person riding a quadruped robot, a fun and powerful demonstration of just how far legged robotics has come. This is the X20 robot from DEEP Robotics. While it's not primarily designed to be ridden, this demonstration is a brilliant way to showcase the platform's core capabilities: a powerful payload capacity, advanced motion control, and incredible stability even with a dynamic load—like a human rider. Beyond the "wow" factor, this is what robust, real-world robotics looks like. The same strength and balance that allows the X20 to carry a person is what will enable it to carry heavy sensor payloads for industrial inspections, navigate disaster sites with supplies, or perform other demanding tasks in complex environments. It's a fun look at some very serious technology. #Robotics #Quadruped #DEEPRobotics #Innovation #Engineering #AI #FutureOfMobility

🚶♂️ Real-time Person Following — Now Live on #DynoAtman Watching our earlier-built algorithms come alive in real-world action is a truly special feeling. A huge appreciation to our brilliant engineers for making this possible! - Our robots can now follow you seamlessly, even through occlusions and dynamic environments. - Fused with stereo camera depth data, this feature enables smooth, responsive, and intelligent motion control, perfectly aligned to a person’s position in real time. #robotics #ai #autonomy #engineering #visionsystems #industrial

Westlake Robotics’ GAE just demoed real-time, low-latency control with arbitrary actions across tasks and hardware. Big step for generalist robot skills and responsive manipulation. What use case would you try first—factory, home, or research? #WestlakeRobotics #GAE #AI #Robotics #Humanoid #EmbodiedAI

Excited to share my latest ROS1 project — "Service Client for TurtleBot3" 🤖 In this demo, the TurtleBot3 traces a circular path, where: 🔴 The first half of the circle is traced in red, and 🔵 The remaining half is traced in blue. This project demonstrates: ✅ Implementation of a ROS1 service–client architecture ✅ Control of 2D robot motion using custom service calls ✅ Integration of visual markers and color-based path tracking Working with ROS (Robot Operating System) continues to enhance my understanding of robotics communication frameworks and motion control. #ROS #TurtleBot3 #Robotics #ROS1 #ServiceClient #RobotProgramming

Here’s a demo using my KUKA KR210 setup in ROS 2 and Gazebo. The robot model is integrated with MoveIt for kinematics, motion planning, and trajectory execution. In this example, the KR210 performs a complete pick-and-place routine, moving an object into the trash. 🔗 GitHub Repository: https://lnkd.in/e_thjVVu #ROS2 #Gazebo #MoveIt #Robotics #MotionPlanning #Automation #OpenSource #Engineering