Remediation is a field within environmental engineering that aims to remove various different contaminants from outdoor environments and consequently remediating the issue of contamination with a certain area. Due to the hazardous nature of contaminants and the locations they may exist in, there exists a demand for mobile robots that may inspect and characterize areas of contaminants. Swift and efficient characterization of contaminated areas can save remediation professionals time and money while also preserving the health of those who might otherwise be characterizing an area. My role in this project was to integrate all the subsystem together into a full-autonomy robot stack.

This environmental sampling robot utilizes a Portable X-Ray Fluorescence (PXRF) sensor in order to obtain quantitative data about elemental contamination. Additionally, a tool arm on the back of the robot allows certain tools to prepare sites to be analyzed by either removing vegetation or a small amount of top soil. A lidar as well as a  GPS antenna are installed for vision and localization purposes. Lastly, a swiveling camera is installed on the robot for inspection in remote environments. All of these hardware components are integrated together through the use of the Robot Operating System to allow the robot full autonomy in performing its task of characterizing analytes.

Preliminary results at various industrial sites show the abilities of this robot in characterizing contaminants in soils over a large region. In a singular use case of evaluating nine points within a region, only a singular intervention was necessary. The robot excelled in its obstacle avoidance while refraining from any terminal collisions that might cause hardware damage. Also, the PXRF was shown to take very accurate measurements of soil contaminants and therefore create an distribution of all contaminants with the affected region.

In conclusion, results show that robot autonomy can aid in effectively characterizing a region of contaminants without the need for a human operator. However, further work is required in refining the autonomy stack to work within a greater scope of chaotic outdoor environments. Further work would involve refining the navigation software stack to better handle travelling through extremely dense vegetation.