Macropod

In an already completed project a test platform was constructed that enabled the acquisition of different sensor data. We developed a special software to visualize the sensor data. Furthermore, algorithms have been developed to determine the negotiability of the environment.

Environment perception plays an important role for the autonomous system. In order to classify the environment, the data of multiple sensors are fused. In the case of the cameras, an extrinsic calibration to the coordinate system of the 3D laser range finder was performed.

The surrounding terrain needs to be analyzed thoroughly according to the negotiability of the terrain surface. Therefore a Markov Random Field was implemented to classify the environment in real-time based on the fused camera-lidar data.

Another objective of the project is the recognition and processing of dynamic changes in the environment of the autonomous vehicle. Dynamic objects, such as vehicles or pedestrians, require an enhanced approach. Their movement needs to be detected early and reliable and needs to be tracked over time. Thereby a collision with the own trajectory can be detected and avoided successfully.

Path planning in unstructured environments is a challenging task for the autonomous system. The robot has to react quickly to changes, needs to be able to incorporate with uncertainties and errors and has to regard the terrain negotiability in real-time. We developed an new form of spline-based outdoor navigation, the spline templates, in order to solve this task for Macropod. This will also include the creation of a global path planning map in the future, with which the current sensor data can be compared to and which can be used for long term navigation.