Weed classification is a fundamental perception task for agricultural robots and an essential enabler of precision and sustainable farming. Existing solutions often rely on high-power edge computing platforms, which limit long-term autonomous operation in Internet of Things (IoT) environments. Meanwhile, the computational complexity of high-accuracy deep learning models hinders their deployment on resource-constrained micro-controllers (MCUs), a critical component of IoT edge nodes. To address these challenges, this paper proposes a TinyML-enabled energy-efficient IoT framework for on-device weed classification, integrating a novel Three-Dimensional Alignment Knowledge Distillation (TDA-KD) strategy with a lightweight multi-layer dilated-convolution student network. The framework enhances knowledge transfer by jointly aligning (i) individual predictions, (ii) inter-sample correlations, and (iii) class semantics, further strengthened through a multi-temperature calibration mechanism. Experimental results on the DeepWeeds and 4Weeds datasets demonstrate that the proposed student model achieves over 95% classification accuracy with only 240K parameters and 87.51 MFLOPs. The model is successfully deployed on an OpenMV H7 Plus board with an STM32H7 MCU, requiring just 105.68 KB Flash memory and achieving an inference time of 378.3 ms with 510.7 mJ energy consumption per sample. A runtime analysis on the Vitirover horticultural robot shows that, compared with a Jetson Nano-based implementation, the proposed IoT pipeline extends operational time by approximately 30.5%. These results highlight the feasibility of deploying high-accuracy weed classification directly on ultra-low-power IoT devices, thereby significantly enhancing the autonomy, energy efficiency, and scalability of agricultural robots.

Accurate long-term place recognition is crucial for agricultural robots operating in unstructured environments. However, in the challenging scene of orchard with high-frequency repetitive features, traditional LiDAR-based localization methods relying on geometric features prove to be inadequate. To address this challenge, we propose TriLoc-NetVLAD, a novel LiDAR-based long-term place recognition approach designed to handle the repetitive and ambiguous features of orchards. This approach initially fuses the point cloud density, height and spatial information to encode unordered 3D point clouds into a spatial context descriptor. then channel selection strategy based on descriptor’s sublayer similarity between query and its corresponding positive and negative samples is proposed to amplify the differences in environmental features. Finally, we use a Triplet Network to extract local features, encompassing both high-dimensional and low-dimensional information. These local features are then cascaded through NetVLAD layer to form a global descriptor. Furthermore, we have built a cross-seasonal orchard dataset to evaluate the performance of our place recognition method. The experiment results demonstrate the advantageous localization performances of the proposed place recognition algorithm over the existing methods.

The emergence of harvesting robotics offers a promising solution to the issue of limited agricultural labor resources and the increasing demand for fruits. Despite notable advancements in the field of harvesting robotics, the utilization of such technology in orchards is still limited. The key challenge for harvesting robots is to improve the operational efficiency. Taking into account inner-arm conflicts, couplings of DoFs, and the dynamic tasks, we propose a task planning strategy for a harvesting robot with four arms in this paper. The proposed method employs a Markov game framework to formulate the four-arm robotic harvesting task, which avoids the computational complexity of solving an NP-hard scheduling problem. Furthermore, a multi-agent reinforcement learning (MARL) structure with a fully centralized collaboration protocol is used to train a MARL-based task planning network. Several simulations and orchard experiments are conducted to validate the effectiveness of the proposed method for a multi-arm harvesting robot in comparison with the existing method.

RGB-D cameras have been successfully used for indoor High-ThroughPut Phenotyping (HTPP). However, their capability and feasibility for in-field HTPP applications still need to be evaluated. To solve the problem, we evaluate the depth-ranging performances of a consumer-level RGB-D camera (RealSense D435i) under in-field scenarios. First, we focus on determining their optimal ranging areas for different crop organs. Second, based on the evaluation results, we analyze the influences of light intensity on depth measurements and propose a brightness-and-distance based Support Vector Regression Strategy, to compensate the ranging error. Finally, we give an intuitive accuracy ranking diagram for RealSense D435i under natural lighting intensities. Experimental results show that: 1) RealSense D435i has good ranging performances on in-field HTPP. 2) Our error compensation model can effectively reduce the influences of lighting intensity and target distance.