Cognibot

Traditional robots struggle with dynamic, unstructured environments because they separate perception ("what do I see?") from action ("what should I do?"). Cognitive science suggests biological agents succeed because they use shared representations—the same neural codes for seeing and acting. This project explores whether robots can achieve similar adaptability.

---

Table of Contents

• Overview
• Tech Stack
• Installation
• Usage
• Current Capabilities
• References
• Contributing

---

Overview

Cognibot is a cognitive robotics project exploring the concept of shared representations in embodied artificial agents. Unlike traditional robotics architectures that separate perception, planning, and action into distinct modules, this project explores how shared representations can unify perception and action planning.

The project is a proof of concept prototype implementation, inspired by the Theory of Event Coding (TEC) (Hommel et al., 2001) and HiTEC computational model (Haazebroek et al., 2011) principles in a simulated differential drive robot using Unity for physics simulation and ROS 2 for cognitive architecture implementation.

---

Tech Stack

• Simulation: Unity 6.0 (6000.0.54f1) LTS with ROS TCP Connector, URDF Importer
• Middleware: ROS 2 Jazzy on Ubuntu 24.04
• Languages: C# (Unity), Python (ROS 2), XML (URDF/Launch files)

---

Installation

Prerequisites

• Unity 6.0 LTS (or compatible) with ROS TCP Connector package (See: ROS TCP Connector )
• ROS 2 Jazzy (installed on Ubuntu 24.04)

Clone Repository

git clone https://github.com/rasika-lokhande/cognibot.git 

Build ROS 2 Workspace

cd cognibot/ros2_ws
colcon build 
source install/setup.bash

---

Usage

In Unity

1. Open the cognibot_sim Unity project
2. Configure ROS Connection GameObject
   • Set ROS IP to your ROS 2 machine's IP address

Running simulation

1. Press Play in Unity to start the simulation

2. Launch the ROS 2 nodes to connect Unity with the cognitive architecture

# Terminal 1: Launch the cognitive architecture

cd cognibot/ros2_ws
source install/setup.bash
ros2 launch cognibot_bringup cognibot_unity.launch.xml 

# This launches the ROS 2 nodes, connects to Unity, and opens Rviz for visualization

3. Monitor feature activations

# Terminal 2: Monitor feature activations
ros2 topic echo /feature_codes

---

Project Structure

cognibot/
├── ros2_ws/                                    # ROS2 workspace
│   ├── src/
│   │   ├── cognibot_interfaces/                # Custom message definitions for feature codes
│   │   ├── common_coding/                      # Common coding implementation
│   │   │   ├── src/
│   │   │   │   ├── lidar_processor.py          # Spatial feature extraction and activation
│   │   │   │   └── action_executor.py          # Execute motor commands based on feature codes
│   │   ├── cognibot_description/               # URDF, robot description
│   │   └── cognibot_bringup/                   # Launch files and configurations
├── cognibot_sim/                               # Unity simulation environment
│   ├── Assets/
│   │   ├── Scripts/
│   │   │   ├── LidarSensor.cs                  # Configurable LIDAR implementation
│   │   │   └── AGVController.cs                 # Robot movement controller
├── docs/                                        # Documentation
│   ├── ideas.md                                 # Project ideas and concepts
└── README.md

---

Current Capabilities

• Sensorimotor Integration: Direct coupling between LIDAR perception and motor actions
• Feature driven navigation: A simple mobile robot navigates simple environments using spatial feature codes

✅ Completed

• ROS 2-Unity integration: Real-time sensor/actuator communication
• Basic URDF model: Robot description for Unity simulation
• Basic robot controller: Differential drive robot with Unity physics
• LIDAR sensor simulation: Configurable LIDAR with Unity
• Feature code generation: Spatial feature extraction from LIDAR data
• Feature activation: Real-time feature activation logging
• Motor command execution: Basic motor commands based on feature codes
• ROS 2 message definitions: Custom messages for feature codes
• Basic cognitive architecture: ROS 2 nodes for feature processing and action execution
• Reactive navigation: Obstacle avoidance using weighted feature activation

🔄 In Progress

• HiTEC neural dynamics: Competition, lateral inhibition, activation
• Task context system: Goal-directed attention and feature modulation

📋 Next steps

• Motor babbling phase: Random exploration for action-effect discovery
• Ideomotor learning: Weight adaptation through action-effect associations

---

References

Key papers informing this implementation:

• Haazebroek, P., van Dantzig, S., & Hommel, B. (2011). A computational model of perception and action for cognitive robotics. Cognitive Processing, 12(4), 355-365.
• Hommel, B., Müsseler, J., Aschersleben, G., & Prinz, W. (2001). The theory of event coding (TEC): A framework for perception and action planning. Behavioral and Brain Sciences, 24(5), 849-878.

---

Contributing

This is an active cognitive robotics experimental project. Contributions, collaborations, and feedback are welcome!

---

"The same neural codes that help you perceive a coffee cup's handle also prepare your hand to grasp it. Can we build robots that work the same way?"