The following abstracts will be presented at the workshop:

Building a Robust, Autonomous Pest-Control Vehicle for Real-World Agricultural Deployment

Bennett Huang 

Globally, pests destroy over 40% of crops an nually. Current application methods include inefficient hand spraying and the use of machines that conduct wasteful mass spraying which pollutes the environment and harms plants. In this paper, we present an autonomous pest-control vehicle that applies pesticides only when pests are detected, conducting targeted pest-control in real-world agricultural environments. We first designed a custom detection pipeline using SIFT-based keypoints and a YOLOv5-inspired network to recognize small pests accurately in real-time. Then, once a pest is detected, the vehicle will turn a cam piece that presses down a spray nozzle, precisely spraying a small amount of pesticides. The spraying mechanism is attached to two linear actuators, allowing it to reach varying plant heights. In our tests, the vehicle’s precise application yielded similar results compared to traditional mass-spraying while reducing pesticide use by over 90%.

Towards the Use of Wood for the Design of Biodegradable Manipulators

Luis Maldonado

Biodegradable robots for environmental monitoring aim to perform their task and leave no traces once it is done. However, bio-source materials often exhibit large variation of their mechanical properties that hinders the reliability of the robotic application. This abstract addresses this problem by performing a sensitivity analysis of a planar RFRFR continuum parallel robot made with wooden segments. 

BeeRootBot: Biodegradable Robotic Probe for Subterranean Environment Monitoring and Rehabilitation

Sachin Sachin, Barbara Mazzolai

This study presents a minimally invasive robotic  probe inspired by plant root growth, designed for subsoil exploration and future ecosystem monitoring and intervention.  The bioinspired probe advances in soil by mimicking plant root apical growth, creating and consolidating a borehole through  the injection  of a bio-based, biodegradable binder at its tip. This innovative process confines penetration resistance to the tip while generating a hollow tubular structure by harnessing in-situ local soil. The probe’s penetration is facilitated by a  linear actuator, which can be retracted upon reaching the desired depth, thereby minimizing the environmental dispersion of mechatronic components. This approach not only enhances  the efficiency of subsoil exploration (whether on Earth or in outer space) by reducing penetration force requirements and reliance on exogenous materials, but also ensures environmental sustainability by employing biodegradable materials and lowering mechanical footprints. The robotic probe’s design and functionality highlight the potential of bio-inspired technologies to address complex environmental challenges, paving the way for future innovations in ecological research and conservation efforts. This study underscores the importance of integrating biological principles into engineering solutions to develop tools that are both effective and environmentally responsible. 

TreeBotic: Learning-Based Robotics as an Enabler for Biodiversity-Aware Agroforestry Systems

Lennart Troesken

Agroforestry—the integration of trees into cropland—is gaining traction across Europe as a pathway to combine food production with biodiversity conservation and climate resilience. However, the spatial and temporal heterogeneity of these systems poses fundamental challenges for mechanization and field management. At the same time, ecological benefits such as biodiversity conservation do not arise automatically from diversified land use, but require explicit integration into management decisions. This creates a compelling application domain for autonomous robotics: small, adaptive platforms that can navigate heterogeneous environments, perform context-sensitive operations, and incorporate ecological objectives into their decision-making. We present TreeBotic, a research framework that addresses these challenges through three tightly coupled components: (i) intervention-based policy learning that leverages human demonstrations and sparse corrections for adaptive autonomy, (ii) Robot Biodiversity Indicators (RBI)—spatially referenced environmental features that embed ecological objectives as actionable variables into planning and control, and (iii) a coupled robotic–ecological digital twin for long-horizon strategy evaluation under environmental variability. Agroforestry serves as a structured yet dynamic testbed, with transfer potential to other ecologically sensitive environments.

Enabling Environmental Monitoring and Animal-Robot Interaction with a Perciform-Inspired Robotic Fish

Gianluca Manduca, Cesare Stefanini, and Donato Romano

Biomimetic robotic fish are increasingly investigated for underwater monitoring and animal–robot interaction  (ARI), where performance strongly relies on the ability to reproduce biologically consistent morphology and motion. However, most ARI studies with fish remain confined to controlled laboratory settings, limiting their ecological relevance. In this work, we introduce a perciform-inspired robotic fish developed through a morphology-driven design to enable minimally invasive interaction in field conditions. The system adopts an underactuated magnetic transmission, combining one active and one passive joint to produce a discrete traveling wave along the body. Experimental results demonstrate stable and controlled locomotion, achieving forward speeds above 0.3 body lengths per second and a minimum Cost of Transport of approximately 14. The platform also exhibits effective steering capabilities and reliable operation across both controlled environments and real-world marine scenarios. These findings emphasize the importance of morphology and magnetic underactuation in achieving compact, robust, and biologically consistent robotic fish, supporting future applications in in situ ARI and ecological monitoring.

Information-Theoretic Analysis of Termite Caste Interactions Toward Bio-Inspired Swarm Robotics for Environmental Monitoring

Gianluca Manduca, Cesare Stefanini, and Donato Romano

Social insects offer powerful biological models for designing distributed robotic systems capable of operating in complex and unstructured environments. Termites,in particular, coordinate collective behaviors through local interactions among specialized castes, providing inspiration for swarm robotics, autonomous sensor networks, and environmental monitoring systems. Here, we analyzed caste-specific interaction dynamics in Reticulitermes lucifugus using a transfer entropy framework. Worker–worker, soldier–soldier, and worker–soldier pairings were recorded and tracked through computer vision. Occupied area and speed were used as behavioral proxies to quantify the amount, direction, and timing of information transfer. Results show that worker–worker pairs exhibited the highest total transfer entropy, reaching 1.04 bits for occupied area and 0.25 bits for speed, compared with lower values in soldier–soldier and mixed pairs. Mixed pairs also showed directional asymmetry, with stronger information flow from soldiers to workers. These findings reveal caste-specific coordination rules that may inform the design of decentralized robotic systems for environmental assessment, where local interactions must support robust collective behavior without centralized control.

Satellite-Anchored Aerial Robotic Sensing for Field Scale Soil Moisture Mapping

Hassan Jaleel

Reliable field-scale soil moisture information is essential for sustainable agricultural water management, partic ularly in irrigated, water-stressed, and data-scarce smallholder farming systems. Effective soil moisture monitoring is also important in conservation agriculture and agroforestry, where improved water management can support carbon sequestra tion and reduce greenhouse gas emissions. However, existing satellite-based soil moisture products are typically too coarse for farm-level decision-making, while in-situ soil moisture sensors provide only point-scale measurements and require substantial deployment costs to cover large agricultural fields. To ad dress these limitations, this work presents a satellite-anchored aerial robotic sensing framework that integrates UAV-based multispectral surveys, satellite-derived soil moisture retrievals, and IoT-based in-situ measurements to generate high-resolution soil moisture maps for farm-level decision support. A key component of the framework is a virtual Short Wave Infra Red (SWIR) sensing module that estimates shortwave infrared reflectance from UAV visible and near-infrared bands, enabling adaptation of the Optical TRApezoid Model to lightweight UAV payloads without specialized SWIR hardware. By bridging the spatial gap between satellite and field observations and the spectral gap between satellite sensors and UAV payloads, the framework offers a perception layer for aerial robotics platform towards a resource-efficient agricultural water management. 

Sustainability robotics expanding the capability of polar research vessels

Luca Romanello

Rapid climate-driven changes in Arctic aquatic ecosystems require new approaches for environmental sensing in regions that remain difficult, costly, and hazardous to access. Large research vessels provide extensive instrumentation and range, but are often unable to safely operate in shallow waters, narrow fjords, and iceberg-dense areas, while also introducing significant operational and ecological footprints. This work presents a vision for sustainable polar environmental robotics that extends the capabilities of small research vessels through distributed aerial–aquatic systems. We report lessons from Greenland field deployments using the MEDUSA aerial– aquatic platform for sensing and sampling in glacier fjords and iceberg-rich environments, including CTD profiling at the ocean– atmosphere interface. Building on these deployments, we discuss complementary robotic modalities, including ecosystem-aware interaction, low disturbance propulsion such as wind-powered surface systems and bio-inspired underwater locomotion, and distributed low footprint sensing platforms, that together enable persistent and minimally invasive environmental observation. We argue that future polar monitoring will rely on coordinated ecosystems of heterogeneous robotic platforms operating across air, surface, and underwater domains. 

UAV-Guided Mobile Manipulation for Autonomous Riverside Litter Cleanup

Viacheslav Krilyshyn

Abstract—Plastic and solid waste accumulation in riverside ecosystems is a major driver of aquatic habitat degradation. We present a multi-robot system that divides litter cleanup across an aerial detection platform and a ground mobile manipulator. Detections from the aerial unit are broadcast as world-frame coordinates and consumed by the ground robot, which navigates   to each item and performs a closed-loop grasp using fused wrist- camera and joint-force feedback. The shared vision module is  trained on a custom dataset collected at the Warta river in Poznan, augmented with a large set of product imagery that extends  coverage of the long-tail object category “garbage.” We describe the architecture, dataset construction, and report observations  from initial indoor manipulation trials. The platform is open- sourced as a starting point for outdoor environmental cleanup deployments.