FlexBot Companion Computer Setup

This repository contains the companion computer (Jetson/Ubuntu PC) side of the FlexBot stack. It is designed to work alongside the IMX7 firmware — see the imx7 branch for the embedded side.

The packages covered here provide a minimal framework for:

• Teleoperation — sending UDP commands to the IMX7 and receiving state feedback
• Wheel Odometry — estimating robot motion from encoder data
• State Estimation — fusing odometry and IMU via an EKF to produce a stable odom → base_link transform

Note: The particle_filter and full bringup packages are not part of this guide. Only the packages listed above are needed for a working teleop + odometry setup.

---

Prerequisites

• ROS 2 (Humble or Jazzy)
• robot_localization — for EKF state estimation
• tf2_ros
• rviz2 (optional, for visualization)

Network configuration (adjust to your setup):

• IMX7 IP: Always check 192.168.0.2
• Companion Computer IP (UDP receiver): Set your computers wired connection ip. 192.168.0.20

---

Build the Workspace

cd ~/flex_bot
colcon build --symlink-install
source install/setup.bash

---

Package Overview

flex_bot_teleop

Handles bidirectional UDP communication with the IMX7.

• udp_bridge_node — bridges ROS 2 topics to/from UDP packets exchanged with the IMX7
• teleop_node — converts joystick or keyboard input into velocity commands

Configuration is in config/flex_bot_udp.yaml and config/teleop.yaml. Adjust UDP ports and IP addresses there to match your network.

Launch teleop:

ros2 launch flex_bot_teleop flex_bot.launch.py

---

flex_bot_odometry

Computes wheel odometry from encoder feedback received over UDP from the IMX7.

Publishes odom → base_link as an odometry source (later fused by the EKF).

Configuration is in config/wheel_odom.yaml.

Launch wheel odometry:

ros2 launch flex_bot_odometry wheel_odom.launch.py

To visualize the odometry path:

ros2 run flex_bot_odometry odom_to_path.py

---

State Estimation (flex_bot_bringup — EKF only)

The EKF (via robot_localization) fuses wheel odometry and IMU data to produce a smooth, stable odom → base_link transform.

Configuration is in flex_bot_bringup/config/ekf_imu.yaml. Key settings:

• two_d_mode: true — recommended for ground robots
• Fuse yaw and yaw-rate from IMU; avoid fusing raw linear acceleration until a reliable odometry source is confirmed, as it will cause drift

Launch state estimation:

ros2 launch flex_bot_bringup bringup_state_estimation.launch.py

Or run the EKF node directly:

ros2 run robot_localization ekf_node --ros-args --params-file \
  ~/flex_bot/src/flex_bot_bringup/config/ekf_imu.yaml

---

Static TF Setup

The robot requires static transforms declaring sensor mounting positions relative to base_link. Publish these before starting any other nodes.

base_link → IMU:

ros2 run tf2_ros static_transform_publisher \
  0 0 0 0 0 0 base_link xsens_imu

Replace the zeros with the real mounting offset once measured.

Verify the full TF tree at any time:

ros2 run tf2_tools view_frames

---

TF Chain

The expected transform chain during normal operation:

odom → base_link        # published by robot_localization EKF
         └→ xsens_imu  # static TF

map → odom is not published in this minimal setup. It would require a SLAM or localization node, which is out of scope here.

---

Recommended Launch Order

1. Static TFs
2. Teleop (UDP bridge + command input)
3. Wheel Odometry
4. State Estimation (EKF)

# Terminal 1 — Static TFs
ros2 launch flex_bot_bringup static_tfs.launch.py

# Terminal 2 — Teleop + UDP bridge
ros2 launch flex_bot_teleop flex_bot.launch.py

# Terminal 3 — Wheel odometry
ros2 launch flex_bot_odometry wheel_odom.launch.py

# Terminal 4 — EKF state estimation
ros2 launch flex_bot_bringup bringup_state_estimation.launch.py

---

IMX7 Counterpart

The embedded firmware running on the IMX7 handles low-level motor control, encoder reading, and UDP packet formatting. See the imx7 branch of this repository for that code.

Make sure the UDP IP/port settings in flex_bot_teleop/config/flex_bot_udp.yaml match what is configured on the IMX7 side.