Active Perception for Accurate Object Localization and Navigation

Project Overview

We are building a TurtleBot4 system that can localize a target object (e.g., a box or cylinder) using RGB-D perception and then autonomously move to improve that estimate. Starting from an initial view, the robot computes an object pose estimate along with a confidence score. Based on this estimate, it selects a next-best viewpoint to reduce pose uncertainty through an active perception strategy.

This perception–action loop repeats until the pose estimate reaches a desired accuracy threshold. During this process, the robot plans and navigates to each next-best viewpoint while safely handling static and dynamic obstacles using ROS 2 navigation tools.

Team

Mohammad Nasr

---

• 🤖 I am currently involved in research focused on Robot Learning.
• 🧠🔧 My expertise spans Robot Software Development, Aerial Robots, Dynamics & Vibration, Data Acquisition Systems, Acoustics, Signal Processing, and CAD Design.
• 📫 How to reach me: nasrmohammad661@gmail.com, mnasr3@asu.edu

Vikas Narang

---

• 🤖 I am currently involved in research focused on Robot Learning.
• 🧠🔧 Research Interests: Robot autonomy, mapping, motion planning , controls , data driven controls, simulation.
• 📫 How to reach me: vikasnar@gmail.com, vnarang2@asu.edu

Name Lastname

• Role: Integration
• Affiliation: Arizona State University
• Research Interests: systems integration, ROS2, embedded robotics
• GitHub: https://github.com/username

Repository Contents

• ROS2 packages for perception and navigation
• Pose estimation pipeline
• Simulation and real-robot integration
• Documentation and experiments

Project Goals

• Accurate object localization with RGB-D: estimate the target object’s pose on the ground plane (x, y, θ) in real time.
• Active perception: choose the next robot viewpoint that improves pose accuracy using a confidence/uncertainty metric.
• Autonomous navigation: move between viewpoints and to the final approach pose without teleoperation.
• Integrate perception with robot decision-making

Quick Links

• 📂 Repository: https://github.com/mohammadnsr1/MobileRobots_Active_Perception.git
• 📖 Documentation: docs/
• 🎥 Demo: (add later)