5G mobile networks introduce a new dimension for connecting and operating mobile robots in outdoor environments, leveraging cloud-native and offloading features of 5G networks to enable fully flexible and collaborative cloud robot operations. However, the limited battery life of robots remains a significant obstacle to their effective adoption in real-world exploration scenarios. This paper explores, via field experiments, the potential energy-saving gains of OROS, a joint orchestration of 5G and Robot Operating System (ROS) that coordinates multiple 5G-connected robots both in terms of navigation and sensing, as well as optimizes their cloud-native service resource utilization while minimizing total resource and energy consumption on the robots based on real-time feedback. We designed, implemented and evaluated our proposed OROS in an experimental testbed composed of commercial off-the-shelf robots and a local 5G infrastructure deployed on a campus. The experimental results demonstrated that OROS significantly outperforms state-of-the-art approaches in terms of energy savings by offloading demanding computational tasks to the 5G edge infrastructure and dynamic energy management of on-board sensors (e.g., switching them off when they are not needed). This strategy achieves approximately 15% energy savings on the robots, thereby extending battery life, which in turn allows for longer operating times and better resource utilization.

Event cameras have emerged as a promising sensing modality for autonomous navigation systems, owing to their high temporal resolution, high dynamic range and negligible motion blur. To process the asynchronous temporal event streams from such sensors, recent research has shown that a mix of Artificial Neural Networks (ANNs), Spiking Neural Networks (SNNs) as well as hybrid SNN-ANN algorithms are necessary to achieve high accuracies across a range of perception tasks. However, we observe that executing such workloads on commodity edge platforms which feature heterogeneous processing elements such as CPUs, GPUs and neural accelerators results in inferior performance. This is due to the mismatch between the irregular nature of event streams and diverse characteristics of algorithms on the one hand and the underlying hardware platform on the other. We propose Ev-Edge, a framework that contains three key optimizations to boost the performance of event-based vision systems on edge platforms: (1) An Event2Sparse Frame converter directly transforms raw event streams into sparse frames, enabling the use of sparse libraries with minimal encoding overheads (2) A Dynamic Sparse Frame Aggregator merges sparse frames at runtime by trading off the temporal granularity of events and computational demand thereby improving hardware utilization (3) A Network Mapper maps concurrently executing tasks to different processing elements while also selecting layer precision by considering both compute and communication overheads. On several state-of-art networks for a range of autonomous navigation tasks, Ev-Edge achieves 1.28x-2.05x improvements in latency and 1.23x-2.15x in energy over an all-GPU implementation on the NVIDIA Jetson Xavier AGX platform for single-task execution scenarios. Ev-Edge also achieves 1.43x-1.81x latency improvements over round-robin scheduling methods in multi-task execution scenarios.

As deep neural networks (DNNs) are being applied to a wide range of edge intelligent applications, it is critical for edge inference platforms to have both high-throughput and low-latency at the same time. Such edge platforms with multiple DNN models pose new challenges for scheduler designs. First, each request may have different service level objectives (SLOs) to improve quality of service (QoS). Second, the edge platforms should be able to efficiently schedule multiple heterogeneous DNN models so that system utilization can be improved. To meet these two goals, this paper proposes BCEdge, a novel learning-based scheduling framework that takes adaptive batching and concurrent execution of DNN inference services on edge platforms. We define a utility function to evaluate the trade-off between throughput and latency. The scheduler in BCEdge leverages maximum entropy-based deep reinforcement learning (DRL) to maximize utility by 1) co-optimizing batch size and 2) the number of concurrent models automatically. Our prototype implemented on different edge platforms shows that the proposed BCEdge enhances utility by up to 37.6% on average, compared to state-of-the-art solutions, while satisfying SLOs.

Implementing existing federated learning in massive Internet of Things (IoT) networks faces critical challenges such as imbalanced and statistically heterogeneous data and device diversity. To this end, we propose a semi-federated learning (SemiFL) framework to provide a potential solution for the realization of intelligent IoT. By seamlessly integrating the centralized and federated paradigms, our SemiFL framework shows high scalability in terms of the number of IoT devices even in the presence of computing-limited sensors. Furthermore, compared to traditional learning approaches, the proposed SemiFL can make better use of distributed data and computing resources, due to the collaborative model training between the edge server and local devices. Simulation results show the effectiveness of our SemiFL framework for massive IoT networks. The code can be found at https://github.com/niwanli/SemiFL_IoT.

As a set of technologies based on edge computing infrastructure and 5G or high-speed connectivity, it is rapidly becoming a reality with many real-world implementations. Statista estimates global IoT spending will hit 1.1 trillion dollars this year, predicting the number of IoT devices worldwide will shoot up from 13 billion currently, to more than 29 billion in 2030. Since it is based on "things", its direct impact on the day-to-day work of senior management or finance departments may seem remote. But IoT can sustain AI (Artificial Intelligence) and machine learning (ML) applications handling multitudes of data flowing from many types of sensors and devices, even when mobile. It opens up huge opportunities for everyone, whether supplying or using services.

NVIDIA expanded the NVIDIA Jetson AI at the Edge platform with an entry-level developer kit priced at just $59, opening the potential of AI and robotics to a new generation of students, educators and hobbyists. The Jetson Nano 2GB Developer Kit is designed for teaching and learning AI by creating hands-on projects in such areas as robotics and intelligent IoT. To support the effort, NVIDIA also announced the availability of free online training and AI-certification programs, which will supplement the many open-source projects, how-tos and videos contributed by thousands of developers in the vibrant Jetson community. "While today's students and engineers are programming computers, in the near future they'll be interacting with, and imparting AI to, robots," said Deepu Talla, vice president and general manager of Edge Computing at NVIDIA. "The new Jetson Nano is the ultimate starter AI computer that allows hands-on learning and experimentation at an incredibly affordable price."

Hailo, the AI chip startup, is teaming with a Japanese systems designer and manufacturing giant Foxconn to develop an AI edge processor aimed at video analytics applications. The edge partnership is based on Foxconn Technology's BOXiedge platform that integrates the Hailo-8 deep learning processor with a parallel processor from Japanese system-on-chip designer Socienext Inc. The combination creates a local video management server designed to shift workload processing from the cloud to the edge. Yokohama-based Socionext specializes in low-power edge processors, including the recent release of an AI processor based on deep neural network technology. The approach is said to be "quantized," that is, the device is tuned to specific values such as signals, rather than a continuous range of values.

Developers using the edge computing platform may be interested in a new piece of hardware created by UP based in the Netherlands called UP AI Edge. As the name suggests it adds artificial intelligence and hardware acceleration'ON' the Edge of the Internet of Things. Watch the demonstration video below to learn more about the UP hardware and its specifications. The UP AI Edge platform offers developers and users and ultra-compact form factor, combining three powerful Intel technologies together into one device providing an Intel Apollo Lake SoC, Intel Movidius Myriad 2 VPU and Intel Cyclone 10GX FPGA. "Not long ago the trend has changed from the Internet of Things to Artificial Intelligence. At UP, we didn't only stop at talking about this shift but we began to actually make it! Not forgetting the post-it sized computing platform, UP Core, we've been launching each exciting project successively. With refueled excitement, we bring you UP AI Edge. The first embedded, ultra compact, high performance, low power consumption, Artificial Intelligence platform for Edge computing."

Enterprises and the public sector worldwide are looking for ways to increase security, improve productivity, provide higher levels of service and reduce maintenance costs. Many of them are using IoT technologies to improve their critical business processes or to drive innovation across their product lines. According to MachNation forecasts and the IoT Edge ScoreCard 2018, worldwide IoT application enablement revenue will be $1.8 billion in 2017 growing to $64.6 billion by 2026 at a compound annual growth rate of 49%. The IoT is imminent – and so are the security challenges it will inevitably bring. Get up to speed on IoT security basics and learn how to devise your own IoT security strategy in our new e-guide.