Pedestrian Density Based Path Recognition and Risk Prediction for Autonomous Vehicles

Mokhtari, Kasra; Ayub, Ali; Surendran, Vidullan; Wagner, Alan R.

doi:10.1109/ro-man47096.2020.9223554

Cited by 5 publications

(3 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Including a more dynamic estimation of the fill levels could yield an improvement in route planning quality. Incorporating other complex characteristics of the SEAs such as pedestrian densities can also be incorporated in future research, as this would provide a more dynamic approach in terms of real-time-based route planning for varying pedestrian densities [33]. Also, the position of the litter bins as compared to other neighboring litter bins (for example, with a radius of 10 m) can also play an influential role on the filling rates.…”

Section: Limitations and Outlookmentioning

confidence: 99%

Minimum Energy Utilization Strategy for Fleet of Autonomous Robots in Urban Waste Management

Justo,

Gupta,

Umland

et al. 2023

Robotics

View full text Add to dashboard Cite

Many service robots have to operate in a variety of different Service Event Areas (SEAs). In the case of the waste collection robot MARBLE (Mobile Autonomous Robot for Litter Emptying) every SEA has characteristics like varying area and number of litter bins, with different distances between litter bins and uncertain filling levels of litter bins. Global positions of litter bins and garbage drop-off positions from MARBLEs after reaching their maximum capacity are defined as task-performing waypoints. We provide boundary delimitation for characteristics that describe the SEA. The boundaries interpolate synergy between individual SEAs and the developed algorithms. This helps in determining which algorithm best suits an SEA, dependent on the characteristics. The developed route-planning methodologies are based on vehicle routing with simulated annealing (VRPSA) and knapsack problems (KSPs). VRPSA uses specific weighting based on route permutation operators, initial temperature, and the nearest neighbor approach. The KSP optimizes a route’s given capacity, in this case using smart litter bins (SLBs) information. The game-theory KSP algorithm with SLBs information and the KSP algorithm without SLBs information performs better on SEAs lower than 0.5 km2, and with fewer than 50 litter bins. When the standard deviation of the fill rate of litter bins is ≈10%, the KSP without SLB is preferred, and if the standard deviation is between 25 and 40%, then the game-theory KSP is selected. Finally, the vehicle routing problem outperforms in SEAs with an area of 0.5≤5 km2, 50–450 litter bins, and a fill rate of 10–40%.

show abstract

Section: Limitations and Outlookmentioning

confidence: 99%

Minimum Energy Utilization Strategy for Fleet of Autonomous Robots in Urban Waste Management

Justo,

Gupta,

Umland

et al. 2023

Robotics

View full text Add to dashboard Cite

show abstract

“…One of the main issues with all of these approaches is that they require a large number of labeled training images per class and cannot learn from few examples (FSIL problem). To the best of our knowledge, CBCL [10], [21], [22], [23] is currently the only approach that tackles the FSIL problem. Although CBCL generates reasonable accuracy on FSIL problems, its accuracy depends on having a good, task-specific feature extractor trained on a large dataset.…”

Section: Related Work a Incremental Learning Techniquesmentioning

confidence: 99%

F-SIOL-310: A Robotic Dataset and Benchmark for Few-Shot Incremental Object Learning

Ayub,

Wagner

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Deep learning has achieved remarkable success in object recognition tasks through the availability of large scale datasets like ImageNet. However, deep learning systems suffer from catastrophic forgetting when learning incrementally without replaying old data. For real-world applications, robots also need to incrementally learn new objects. Further, since robots have limited human assistance available, they must learn from only a few examples. However, very few object recognition datasets and benchmarks exist to test incremental learning capability for robotic vision. Further, there is no dataset or benchmark specifically designed for incremental object learning from a few examples. To fill this gap, we present a new dataset termed F-SIOL-310 (Few-Shot Incremental Object Learning) which is specifically captured for testing few-shot incremental object learning capability for robotic vision. We also provide benchmarks and evaluations of 8 incremental learning algorithms on F-SIOL-310 for future comparisons. Our results demonstrate that the few-shot incremental object learning problem for robotic vision is far from being solved.

show abstract

“…Considering the wider field of mobility and traffic control, most of the existing body of works focuses on risk-aware road movement and path planning to avoid collisions with infrastructure, other vehicles, or pedestrians [11], [12]. Additional examples include [13], that describes an architecture for supporting the risk-aware operation of autonomous vehicles; [14] defines safe and hazardous states for autonomous vehicles in an urban area, and [15] introduces an approach for minimising the possibility of collisions with pedestrians. Finally, [16] investigates how risk measures can affect the behaviours and trajectories of autonomous vehicles in an urban environment.…”

Section: Introductionmentioning

confidence: 99%

A Framework for Risk-Aware Routing of Connected Vehicles via Artificial Intelligence

Cardellini,

Dodaro,

Maratea

et al. 2023

2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC)

View full text Add to dashboard Cite

The advent of Connected Autonomous Vehicles can enable the use of Artificial Intelligence (AI) techniques to support urban traffic controllers in extending their control capabilities with the ability to distribute vehicles in a urban region. Vehicles can communicate their destination, and receive an optimised route by traffic controllers. While the benefits of traffic routing are clear, it is also clear that re-routing has the potential to increase risks for vehicles' and passengers' safety due to environmental or urban factors. There is however a lack of work in the area of risk-aware routing.To fill the above-mentioned gap, we introduce a framework to incorporate risk-awareness in the vehicle routing process. The proposed framework provides a principled structure to define and characterise different classes of risk that can arise in a region, allowing to take them into account when generating routes. We show how this framework can be implemented, and we provide an empirical analysis of its performance on two European urban areas.

show abstract

Pedestrian Density Based Path Recognition and Risk Prediction for Autonomous Vehicles

Cited by 5 publications

References 32 publications

Minimum Energy Utilization Strategy for Fleet of Autonomous Robots in Urban Waste Management

Minimum Energy Utilization Strategy for Fleet of Autonomous Robots in Urban Waste Management

F-SIOL-310: A Robotic Dataset and Benchmark for Few-Shot Incremental Object Learning

A Framework for Risk-Aware Routing of Connected Vehicles via Artificial Intelligence

Contact Info

Product

Resources

About