Survival-Oriented Reinforcement Learning Model: An Effcient and Robust Deep Reinforcement Learning Algorithm for Autonomous Driving Problem

Ye, Changkun; Ma, Hui; Zhang, Xiaoqin; Zhang, Kai; You, Shaodi

doi:10.1007/978-3-319-71589-6_36

Cited by 13 publications

(5 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In multifidelity reinforcement learning; the policies for low fidelity simulators were transferred to high fidelity simulators for exploration as heuristics to find optimal policies with less data [28]. A constrained Markov Decision Process (MDP) called survival oriented RL which takes survival (Negative avoidance) as the first priority rather than maximizing reward is considered in [29] for ensuring safety. Multi-Agent RL has been used for high level strategic decision making such as overtaking, following vehicles using dynamic coordination graphs [30].…”

Section: Modern Approachesmentioning

confidence: 99%

Deep 3D Dynamic Object Detection towards Successful and Safe Navigation for Full Autonomous Driving

Wijesekara¹

2022

Preprint

View full text Add to dashboard Cite

Infractions other than collisions are also a crucial factor in Autonomous driving since other infractions can result in an Accident. Most of the existing work has been tested for Navigation, collisions and least tested for other infractions such as off-road driving and not obeying road signs. We present an imitation learning based Base Model called CILDO and an optimized model of the base model optimized using an and additional traffic light detection branch and Deep Deterministic Policy Gradient based Reinforcement Learning called CILDOLI-RL. The proposed base model is designed to detect dynamic objects by developing the vision for semantic features, depth and motion which is crucial in Autonomous driving. The CILDO-RL model presented in this paper achieves highest score for the newly introduced No-OTHERINFRACTION benchmark ensuring safe autonomous driving while the base CILDO model achieves the best performance in Navigation under Urban or rural dense traffic environments. Further, this work makes a comparison on the computational complexities of the driving models.<br>

show abstract

Section: Modern Approachesmentioning

confidence: 99%

Deep 3D Dynamic Object Detection towards Successful and Safe Navigation for Full Autonomous Driving

Wijesekara¹

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Outside of medicine, (Ye et al, 2017) proposed a method, which aims to optimize the cumulative rewards in a constrained MDP, with a negative avoidance constraint. Their method uses a Negative Avoidance Function (NAF), which plays a role similar to a hazard function.…”

Section: Related Workmentioning

confidence: 99%

Reinforcement Learning For Survival: A Clinically Motivated Method For Critically Ill Patients

Nanayakkara¹

2022

Preprint

View full text Add to dashboard Cite

There has been considerable interest in leveraging RL and stochastic control methods to learn optimal treatment strategies for critically ill patients, directly from observational data. However, there is significant ambiguity on the control objective and on the best reward choice for the standard RL objective. In this work, we propose a clinically motivated control objective for critically ill patients, for which the value functions have a simple medical interpretation. Further, we present theoretical results and adapt our method to a practical Deep RL algorithm, which can be used alongside any value based Deep RL method. We experiment on a large sepsis cohort and show that our method produces results consistent with clinical knowledge.

show abstract

“…The external models are not limited in RL policies. For example, danger index can be predicted with Negative-Avoidance Function [35] and intrinsic fear is approximated by a supervised learning model [5], which all can intervene an original action and improve the safety. Researchers [36], [37], [38] also have investigated reset policies, which are trained jointly with the task policy to gradually expand recoverable regions.…”

Section: B Conservative Explorationmentioning

confidence: 99%

Improving Safety in Deep Reinforcement Learning using Unsupervised Action Planning

Hsu¹,

Huang²,

Ha³

2021

Preprint

View full text Add to dashboard Cite

One of the key challenges to deep reinforcement learning (deep RL) is to ensure safety at both training and testing phases. In this work, we propose a novel technique of unsupervised action planning to improve the safety of onpolicy reinforcement learning algorithms, such as trust region policy optimization (TRPO) or proximal policy optimization (PPO). We design our safety-aware reinforcement learning by storing all the history of "recovery" actions that rescue the agent from dangerous situations into a separate "safety" buffer and finding the best recovery action when the agent encounters similar states. Because this functionality requires the algorithm to query similar states, we implement the proposed safety mechanism using an unsupervised learning algorithm, k-means clustering. We evaluate the proposed algorithm on six robotic control tasks that cover navigation and manipulation. Our results show that the proposed safety RL algorithm can achieve higher rewards compared with multiple baselines in both discrete and continuous control problems. The supplemental video can be found at: https://youtu.be/AFTeWSohILo.

show abstract

Survival-Oriented Reinforcement Learning Model: An Effcient and Robust Deep Reinforcement Learning Algorithm for Autonomous Driving Problem

Cited by 13 publications

References 6 publications

Deep 3D Dynamic Object Detection towards Successful and Safe Navigation for Full Autonomous Driving

Deep 3D Dynamic Object Detection towards Successful and Safe Navigation for Full Autonomous Driving

Reinforcement Learning For Survival: A Clinically Motivated Method For Critically Ill Patients

Improving Safety in Deep Reinforcement Learning using Unsupervised Action Planning

Contact Info

Product

Resources

About