Archaeological fieldwork often begins with time-consuming surveys where researchers manually search for artifacts, such as pottery shards, across large terrain, this process is labor-intensive. The Archebot rover is an autonomous ground rover designed for archaeological field research. Its primary purpose is to assist archaeologists in the early-stage survey of excavation sites by scanning and identifying pottery shards scattered across the surface of the designated area. Utilizing a user-defined area map, the Archebot navigates across the terrain. As it moves, it uses an onboard computer vision model trained to detect pottery shards in real time. The system processes video and image data from ground-facing cameras to identify potential artifacts lying on or near the surface. When a potential shard is detected, Archebot will log the GPS coordinates of the finding, capture and save the image evidence, and flag the location for future retrieval. Upon completing the scan of the assigned area, the rover generates a visual report that maps all detected shard locations, allowing archaeologists to locate and collect the findings for further analysis.
Our client, Tuna Kalaycı, is an Assistant professor at the department of Archaeological Sciences of Leiden University. He has degrees in Statistics, Settlement Archaeology, and Anthropology. Here his main interests are in remote sensing, data analysis, and modeling. He also aims to positively change old processes with modern concepts like digitization and machine automation. The project was developed in close collaboration with him to ensure the solution addressed real-world challenges encountered during archaeological surveys. Communication with the client was excellent. We maintained regular contact throughout the project, receiving valuable insights. This collaboration helped ensure the development stayed aligned with the client's practical needs and expectations.
Our team worked in an agile workflow, assigning clear roles to ensure effictiveness and productivity. Key roles include: Scrum master, who was responsible for tracking progress and organizing meetings. Product Owner, who acted as the main point of contact with the client and ensured the prodcut aligned with user needs. The rest of the team primarily focused on the development of the rover. We divided the work into functional components, hardware, software, and computer vision, based on each team member's strengths. We held regular meetings to synchronize development, used GitHub for version control and communication, and conducted bi-weekly sprint reviews to assess progress and re-prioritize tasks as needed. One of the main challenges we faced was integrating the hardware components such as the GPS and depth camera, which initially caused frequent errors. We also had to fine-tune the object detection system and improve the path-following algorithm, as their initial performance was not optimal. As a team, we are most proud of delivering a functional prototype that met the clients expectations. Our ability to combine hardware, software, and AI into a practical tool for archaeology, and to do so within our project timeline, is something we're proud of.