eva_behavior

Initial behaviour tree for Eva. Current verion designed to work with Dmitroid robot and current nodes already developed.

Pre-requisites

The Owyl Behavior Tree package is required. To install:

1. git clone git@github.com:eykd/owyl.git

2. sudo python setup.py install

Co-requisites

This is meant to be used in conjuction with several ROS nodes:

• blender_api_msgs defines the robot head controlling messages.
• robo_blender which implements a movable robot head.
• [pi_vision]((http://github.com/hansonrobotics/pi_vision) which provides visual data to the perception node.

Don't forget to:

• catkin build
• source devel/setup.bash

Running

To run, start the robo_blender, perception, pi_vision and uvc_cam nodes. Then

rosrun eva_behavior main.py

Then, turn on behaviors:

rostopic  pub --once /behavior_switch std_msgs/String btree_on

This will cause the system to play through the set of scripted behaviors, which includes exprsssing a variety of emotions, looking at the various faces as they become visible, and so on.

ROS Messaging API

Topics subscribed :

• behavior_switch (std_msgs/String). Values: ("btree_on", "btree_off"). By default the behaviour tree is off and it needs to receive btree_on to start.
• tracking_event (eva_behavior/event). Event received from the perception nodes, currently only pi_vision.

Topics published

• cmd_blendermode (std_msgs/String). Blender mode used for tracking: Dummy (do nothing), TrackDev(tracking objects), LookAround (looking for attention).

• /blender_api/set_face_target (blender_api_msg/Target). Sends location that the robot head should face and look at.

• /blender_api/set_gaze_target (blender_api_msg/Target). Sends location that the robot eyes should look at (without moving the head).

Cookbook

Below follows a list of examples and demos.

Basic random emotion cycling

Start up one of the blender heads, then:

rosrun eva_behavior main.py

Turn on behaviors:

rostopic  pub --once /behavior_switch std_msgs/String btree_on

The head should now be cycling through a set of facial expressions.

Face Tracking

As above, but also start uvc_cam and pi_vision. Note that uvc_cam won't automatically show up in the ROS_PACKAGE_PATH environment variable; it must be added by hand.

export ROS_PACKAGE_PATH=/path/to/uvc_cam:$ROS_PACKAGE_PATH
roslaunch ros2opencv uvc_cam.launch device:=/dev/video1
roslaunch pi_face_tracker face_tracker_uvc_cam.launch