Network offloading policies for cloud robotics: a learning-based approach

Chinchali, Sandeep; Sharma, Anita; Harrison, J. Michael; Elhafsi, Amine; Kang, Daniel; Pergament, Evgenya; Cidon, Eyal; Katti, Sachin; Pavone, Marco

doi:10.1007/s10514-021-09987-4

Cited by 41 publications

(11 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Engineers must handle sensor data correctly to overcome such constraints and utilize the benefits of wireless cellular communication. Many studies research this topic, such as [ 10 ].…”

Section: Uav As Uementioning

confidence: 99%

“…Therefore, every engineer must carefully assess the capability of the current wireless communication. Then, one has to design a corresponding sensor data handling mechanism to ensure that the application stays within the bounds of the offered communication while making as much use of the available sensor data as possible [ 7 , 10 ].…”

Section: Introductionmentioning

confidence: 99%

“…Here it is important to note that offloading computationally heavy processes from the UAV to the cloud has many benefits. Characteristic examples of such processes are the execution of intensive image analysis algorithms that would require extensive onboard computations that many small-scale UAVs or robots can not comprehend [ 10 ]. By offloading computationally heavy processes, less advanced onboard processors can be sufficient, which in turn may reduce the cost of the UAV.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Performance of Sensor Data Process Offloading on 5G-Enabled UAVs

Damigos

Lindgren

Sandberg

et al. 2023

Sensors

View full text Add to dashboard Cite

Recently, unmanned aerial vehicle (UAV)-oriented applications have been growing worldwide. Thus, there is a strong interest in using UAVs for applications requiring wide-area connectivity coverage. Such applications might be power line inspection, road inspection, offshore site monitoring, wind turbine inspections, and others. The utilization of cellular networks, such as the fifth-generation (5G) technology, is often considered to meet the requirement of wide-area connectivity. This study quantifies the performance of 5G-enabled UAVs when sensor data throughput requirements are within the 5G network’s capability and when throughput requirements significantly exceed the capability of the 5G network, respectively. Our experimental results show that in the first case, the 5G network maintains bounded latency, and the application behaves as expected. In the latter case, the overloading of the 5G network results in increased latency, dropped packets, and overall degradation of the application performance. Our findings show that offloading processes requiring moderate sensor data rates work well, while transmitting all the raw data generated by the UAV’s sensors is not possible. This study highlights and experimentally demonstrates the impact of critical parameters that affect real-life 5G-enabled UAVs that utilize the edge-offloading power of a 5G cellular network.

show abstract

“…Engineers must handle sensor data correctly to overcome such constraints and utilize the benefits of wireless cellular communication. Many studies research this topic, such as [ 10 ].…”

Section: Uav As Uementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Performance of Sensor Data Process Offloading on 5G-Enabled UAVs

Damigos

Lindgren

Sandberg

et al. 2023

Sensors

View full text Add to dashboard Cite

show abstract

“…Conceived as highly scalable systems with access to potentially unlimited resources, they can accelerate both CNN inference and training on dedicated servers, fully or partially taking on the required computational load and thus relegating user-level devices to mere data-entry and result-presentation terminals [21,22]. Nevertheless, this model presents certain limitations in terms of response speed, availability, and security [23][24][25][26][27][28][29][30][31], which is why it might be inadequate in scenarios where system response time must be as short as possible, in austere contexts with limited communication or computational resources, or even in cases where data privacy is a hard requirement. More specifically:…”

Section: Introductionmentioning

confidence: 99%

“…• Increase in latency, i.e., the response time of the system, does not harm detection accuracy, but it can lead to significant degradation of the experience [24]. Offloading processing tasks to a remote machine adds to inference time-that is, trained model execution timethe amount of time devoted to image encoding and transfer, consequently generating the perception among end-users of higher slowness in the whole process [23,25,28]. • Connectivity between a terminal device and the serverside is crucial.…”

Section: Introductionmentioning

confidence: 99%

Optimized convolutional neural network architectures for efficient on-device vision-based object detection

Rodriguez-Conde

Campos

Fdez-Riverola

2021

Neural Comput & Applic

View full text Add to dashboard Cite

Convolutional neural networks have pushed forward image analysis research and computer vision over the last decade, constituting a state-of-the-art approach in object detection today. The design of increasingly deeper and wider architectures has made it possible to achieve unprecedented levels of detection accuracy, albeit at the cost of both a dramatic computational burden and a large memory footprint. In such a context, cloud systems have become a mainstream technological solution due to their tremendous scalability, providing researchers and practitioners with virtually unlimited resources. However, these resources are typically made available as remote services, requiring communication over the network to be accessed, thus compromising the speed of response, availability, and security of the implemented solution. In view of these limitations, the on-device paradigm has emerged as a recent yet widely explored alternative, pursuing more compact and efficient networks to ultimately enable the execution of the derived models directly on resource-constrained client devices. This study provides an up-to-date review of the more relevant scientific research carried out in this vein, circumscribed to the object detection problem. In particular, the paper contributes to the field with a comprehensive architectural overview of both the existing lightweight object detection frameworks targeted to mobile and embedded devices, and the underlying convolutional neural networks that make up their internal structure. More specifically, it addresses the main structural-level strategies used for conceiving the various components of a detection pipeline (i.e., backbone, neck, and head), as well as the most salient techniques proposed for adapting such structures and the resulting architectures to more austere deployment environments. Finally, the study concludes with a discussion of the specific challenges and next steps to be taken to move toward a more convenient accuracy–speed trade-off.

show abstract

The SoS conductor: Orchestrating resources with iterative agent‐based reinforcement learning

Chen,

Heydari

2024

Systems Engineering

View full text Add to dashboard Cite

We introduce a novel resource management approach for Systems of Systems (SoS), utilizing hierarchical deep reinforcement learning, iterating with agent‐based simulation. A key innovation of this method is its ability to balance top‐down SoS management with the autonomy of individual systems. This is achieved by dynamically allocating resources to each system, thereby modifying the range of options they can autonomously choose from. This dynamic option adjustment is a powerful approach to managing the trade‐off between centralized efficiency and decentralized autonomous actions of the systems, enabling the SoS to maintain the systems' autonomy while ensuring efficient SoS governance. The method, validated through a case study, not only demonstrates the potential and efficacy of the learning framework but also reveals how, using this method, minor performance sacrifices can lead to substantial improvements in resource efficiency.

show abstract

Network offloading policies for cloud robotics: a learning-based approach

Cited by 41 publications

References 31 publications

Performance of Sensor Data Process Offloading on 5G-Enabled UAVs

Performance of Sensor Data Process Offloading on 5G-Enabled UAVs

Optimized convolutional neural network architectures for efficient on-device vision-based object detection

The SoS conductor: Orchestrating resources with iterative agent‐based reinforcement learning

Contact Info

Product

Resources

About