Abstract
Long-term operation of robots creates new challenges to Simultaneous Localization and Mapping (SLAM) algorithms. Long-term SLAM algorithms should adapt to recent changes while preserving older states, when dealing with appearance variations (lighting, daytime, weather, or seasonal) or environment reconfiguration. When also operating robots for long periods and trajectory lengths, the map should readjust to environment changes but not grow indefinitely. The map size should depend only on updating the map with new information of interest, not on the operation time or trajectory length. Although several studies in the literature review SLAM algorithms, none of the studies focus on the challenges associated to lifelong SLAM. Thus, this paper presents a systematic literature review on long-term localization and mapping following the Preferred Reporting Items for Systematic reviews and Meta-Analysis guidelines. The review analyzes 142 works covering appearance invariance, modeling the environment dynamics, map size management, multisession, and computational topics such as parallel computing and timing efficiency. The analysis also focus on the experimental data and evaluation metrics commonly used to assess long-term autonomy. Moreover, an overview over the bibliographic data of the 142 records provides analysis in terms of keywords and authorship co-occurrence to identify the terms more used in long-term SLAM and research networks between authors, respectively. Future studies can update this paper thanks to the systematic methodology presented in the review and the public GitHub repository with all the documentation and scripts used during the review process.

Abstract
Robotics and intelligent systems are key technologies to promote efficient and innovative applications in the most diverse domains (industry, healthcare, agriculture, construction, mobility, etc [...]

Abstract

Abstract

Abstract
Purpose: Visual perception enables robots to perceive the environment. Visual data is processed using computer vision algorithms that are usually time-expensive and require powerful devices to process the visual data in real-time, which is unfeasible for open-field robots with limited energy. This work benchmarks the performance of different heterogeneous platforms for object detection in real-time. This research benchmarks three architectures: embedded GPU-Graphical Processing Units (such as NVIDIA Jetson Nano 2 GB and 4 GB, and NVIDIA Jetson TX2), TPU-Tensor Processing Unit (such as Coral Dev Board TPU), and DPU-Deep Learning Processor Unit (such as in AMD-Xilinx ZCU104 Development Board, and AMD-Xilinx Kria KV260 Starter Kit). Methods: The authors used the RetinaNet ResNet-50 fine-tuned using the natural VineSet dataset. After the trained model was converted and compiled for target-specific hardware formats to improve the execution efficiency.Conclusions and Results: The platforms were assessed in terms of performance of the evaluation metrics and efficiency (time of inference). Graphical Processing Units (GPUs) were the slowest devices, running at 3 FPS to 5 FPS, and Field Programmable Gate Arrays (FPGAs) were the fastest devices, running at 14 FPS to 25 FPS. The efficiency of the Tensor Processing Unit (TPU) is irrelevant and similar to NVIDIA Jetson TX2. TPU and GPU are the most power-efficient, consuming about 5 W. The performance differences, in the evaluation metrics, across devices are irrelevant and have an F1 of about 70 % and mean Average Precision (mAP) of about 60 %.