Multimodal robotic localization and autonomous sampling in greenhouses
An advanced autonomous robotic system for efficient plant tissue sampling in greenhouses using multimodal sensor fusion, enhancing localization and mapping in challenging environments. Integrates lidar SLAM, computer vision, and robotic manipulation for improved autonomy.
Overview
The proposed solution is an autonomous robotic system designed to optimize plant tissue sampling in greenhouse environments. By utilizing advanced multimodal sensor fusion techniques, this system enhances the localization and mapping capabilities in challenging conditions characterized by featureless areas and transparent objects. This integration of lidar SLAM, computer vision, and robotic manipulation ensures precise detection, navigation, and handling of plant tissues, significantly boosting the efficiency and autonomy of greenhouse operations.
Technical specifications
Key features:
- Incorporates an intensity-based detection algorithm into a state-of-the-art 2D lidar SLAM system for improved mapping of transparent objects.
- Utilizes learning-based methods for navigating mobile robots in indoor environments.
- Employs multiple sensors, including IMUs, cameras, and lidars, for accurate state estimation and localization of robots.
- Implements semantic segmentation through computer vision to identify diverse plant species and determine plant tissue sampling locations.
- Integrates autonomous manipulation via rigid robotic arms (e.g., UR3) or soft robots for efficient sample collection and placement.
Technology readiness level
This technology is currently at TRL 3, indicating it is in the experimental proof of concept stage. The system has undergone initial validation efforts, with future developments focusing on enhancing the localization, mapping, and autonomous sampling capabilities within actual greenhouse environments.
About University of Maryland
The University of Maryland is the state's flagship public research university in College Park, serving a large student body and offering comprehensive academic and research programs. Industry engages through an adjacent research park, shared prototyping facilities, and dedicated corporate partnership channels, plus joint pathways to a medical campus for clinical and regulatory engagement. Its Washington–Baltimore location places partners near federal agencies, national labs, and prime contractors, while a statewide extension network supports collaboration with companies. Research is backed by competitive federal funding from agencies such as NSF, NIH, DoD, DOE, NASA, and USDA, and coordinated commercialization services advance IP, startup formation, licensing, and industry-sponsored research.
delivered to your inbox.