Follow-me with ADBSCAN and Gesture-based Control on Clearpath Robotics Jackal Robot#

This tutorial demonstrates the Follow-me algorithm along with a gesture recognition network, where the robot follows a target person in real time and responds to state commands through hand gestures. This tutorial uses Clearpath Robotics Jackal robot and one Intel® RealSense™ camera D400 series. This camera provides the point cloud data as input for the Intel®-patented object detection algorithm Adaptive DBSCAN to detect the position of the target person. This camera also provides RGB images to the object detection network responsible for detecting hand gestures for controlling the robot’s start and stop states. This RGB image is passed through a deep learning-based gesture recognition pipeline, called Mediapipe Hands Framework, to detect the gesture category. The motion commands for the robot are published to twist topic based on these two outputs: person’s position and gesture category.

To start, the robot requires two conditions at the same time:

  • The target person will be within the tracking radius of the robot.

    Radius is a reconfigurable parameter in: /opt/ros/jazzy/share/tutorial-follow-me-w-gesture/params/followme_adbscan_RS_params.yaml.

    In ROS 2 Humble, the file is located at the similar directory path.

  • The detected gesture of the target is thumbs up.

Once the starting criteria are met, the robot keeps following the target unless one of the below stopping conditions are triggered:

  • The target moves to a distance greater than the tracking radius of the robot.

    Radius is a reconfigurable parameter in: /opt/ros/jazzy/share/tutorial-follow-me-w-gesture/params/followme_adbscan_RS_params.yaml.

  • The detected gesture is thumbs down.

Getting Started#

Prerequisites#

Complete the get started guide before continuing.

Install the Deb package#

Install ros-jazzy-follow-me-tutorial-w-gesture Deb package from Autonomous Mobile Robot APT repository.

sudo apt update
sudo apt install ros-jazzy-follow-me-tutorial-w-gesture

sudo apt update
sudo apt install ros-humble-follow-me-tutorial-w-gesture

Install Python Modules#

This application uses Mediapipe Hands Framework for hand gesture recognition. Install the following modules as a prerequisite for the framework:

pip3 install mediapipe
pip3 install numpy==1.24.3

Identify serial number of Realsense Camera#

Install the Intel® RealSense™ camera utilities package to easily read the correct serial number:

sudo apt install librealsense2-utils

Check the Serial number:

$rs-enumerate-devices | grep Serial
 Serial Number                 :     108322074411
 Asic Serial Number            :     108223052439

You will use this serial number (and not the ASIC Serial Number) when launching the demo below.

Run Demo with Intel® RealSense™ Camera#

Execute the following script to launch Follow-Me with Gesture on the Clearpath Robotics Jackal robot:

source /opt/ros/jazzy/setup.bash
ros2 launch tutorial_follow_me_w_gesture jackal_gesture_launch.py <Camera Serial Number>

source /opt/ros/humble/setup.bash
ros2 launch tutorial_follow_me_w_gesture jackal_gesture_launch.py <Camera Serial Number>

<Camera Serial Number>: Use the serial number returned when using rs-enumerate-devices. Note that the output of other programs like lsusb might return an incorrect serial number.

After starting the script, the robot should begin searching for trackable objects in its initial detection radius (defaulting to around 0.5m), and then following acquired targets as soon as they provide a thumbs up to the Intel® RealSense™ camera and move from the initial target location.

There are reconfigurable parameters in the /opt/ros/humble/share/tutorial_follow_me_w_gesture/params directory for the Intel® RealSense™ camera (followme_adbscan_RS_params.yaml). You can modify parameters depending on the respective robot, sensor configuration and environments (if required) before running the tutorial. Find a brief description of the parameters in the following list:

  • Lidar_type

    Type of the point cloud sensor. For Intel® RealSense™ camera and LIDAR inputs, the default value is set to RS and 2D, respectively.

  • Lidar_topic

    Name of the topic publishing point cloud data.

  • Verbose

    If this flag is set to True, the locations of the detected target objects will be printed as the screen log.

  • subsample_ratio

    This is the downsampling rate of the original point cloud data. Default value = 15 (i.e. every 15-th data in the original point cloud is sampled and passed to the core ADBSCAN algorithm).

  • x_filter_back

    Point cloud data with x-coordinate > x_filter_back are filtered out (positive x direction lies in front of the robot).

  • y_filter_left, y_filter_right

    Point cloud data with y-coordinate > y_filter_left and y-coordinate < y_filter_right are filtered out (positive y-direction is to the left of robot and vice versa).

  • z_filter

    Point cloud data with z-coordinate < z_filter will be filtered out. This option will be ignored in case of 2D Lidar.

  • Z_based_ground_removal

    Filtering in the z-direction will be applied only if this value is non-zero. This option will be ignored in case of 2D Lidar.

  • base, coeff_1, coeff_2, scale_factor

    These are the coefficients used to calculate adaptive parameters of the ADBSCAN algorithm. These values are pre-computed and recommended to keep unchanged.

  • init_tgt_loc

    This value describes the initial target location. The person needs to be at a distance of init_tgt_loc in front of the robot to initiate the motor.

  • max_dist

    This is the maximum distance that the robot can follow. If the person moves at a distance > max_dist, the robot will stop following.

  • min_dist

    This value describes the safe distance the robot will always maintain with the target person. If the person moves closer than min_dist, the robot stops following.

  • max_linear

    Maximum linear velocity of the robot.

  • max_angular

    Maximum angular velocity of the robot.

  • max_frame_blocked

    The robot will keep following the target for max_frame_blocked number of frames in the event of a temporary occlusion.

  • tracking_radius

    The robot will keep following the target as long as the current target location = previous location +/- tracking_radius.

Troubleshooting#

  • Failed to run the tutorial mentioning permission denied on /dev/dri/render128

    usermod -a -G render $USER

    Note: The machine may need to be restarted after adding the user to a new group.

  • Failed to install Deb package: Please make sure to run sudo apt update before installing the necessary Deb packages.

  • You may stop the demo anytime by pressing ctrl-C.

  • If the robot rotates more than intended at each step, try reducing the parameter max_angular in the parameter file.

  • If the motor controller board does not start, restart the robot.

  • For general robot issues, refer to Troubleshooting.