During the labs we will be using V-REP simulator with a Python remote API and a real hexapod walking robot. This page provides resources which are common for all the labs.

hexapod_vrep_v2.0

Update: Updated version of hexapod_vrep for dealing with noisy position, orientation and laser scanner data
hexapod_vrep_v3.0

hexapod_real_v1.0

The HAL allows to use standardized function calls for interfacing both the simulated and the real robot regardless the particular implementation of the hardware layer. The particular interface is selected by choosing one of two possible includes in the main python script. For interfacing the simulated robot in the V-REP the main python should include in the header:

import hexapod_vrep.robot as rob

import hexapod_real.robot as rob

Hexapod model

Hexapod servos numbering:

scenes_pack_v1.0
Scenes are the simulation main playground. Obstacles, robots and other objects can be placed into the scene and controlled by the simulation. Besides, terrain maps can be imported from image files directly to the scene.

• Plain scene
  • The very basic scene for testing locomotion principles.
• Simple obstacle
  • Scene with only one obstacle to test the locomotion and basic obstacle detection and obstacle avoidance.
• Obstacle avoidance
  • More advanced scene for testing reactive obstacle avoidance.
• Maze 1-3
  • Scenes for testing of motion planning approaches with obstacles imported from the image file.

Obstacles in the scene can be made by adding a primitive shape into the scene by right clicking into the scene and selecting add » Primitive shape. Then we set up the dimensions of the shape by filling in the dialog.

To make the obstacle static (i.e. not influenced by gravity or collisions) but still exerting some collision responses on non-static respondable shapes it is necessary to disable Body is dynamic option in the shape dynamics properties accessed from the Hierarchy manager and Show dynamic properties dialog as it is visualized in the following picture.

It is also necessary to make the obstacle Collidable, Measurable, Renderable and Detectable in the Common properties of the same dialog.

The maze creation is based on importing the height map into the V-REP simulator. As a source, gray scale .jpg or .png image is used. The maze can be imported into V-REP as follows:

1. Prepare the image of the maze preferably in a low resolution (approx. 200×200)
2. Go to File>Import>Heightfield and locate the image
3. Fill in the imported Heightfield dimensions where Z-scaling is the height difference between the black and white colors
4. Move the Heightfield in z-direction so the bottom of it lies under the ResizableFloor
5. Resize the ResizableFloor so it allows the robot to move freely through the whole maze
6. Set up the properties of the Heightfield according to the above described approach. Hence, make the object Collidable, Measurable, Renderable and Detectable.

demo_codes_v1.0
V-REP demos intends to show the basics of controlling the hexapod model in the V-REP simulator.

• Joint position setting demo
  • This demo shows how to precisely control the joint angles of the robot.
• Simple walking demo
  • This demo shows a basic locomotion control of a hexapod robot using the Central Pattern Generator (CPG) neural network.
  • By adjusting variables l and r in the main function, the robot performs locomotion with a different turning radius and speed
• Distance sensor readout demo
  • This demo shows how to interface the simulated distance sensor. Note, in the simulation the sensor is simulated as a ray type laser scanner whereas on the real robot an Asus Xtion Pro RGB-D camera is used to obtain the distance measurements.