Proximal Policy Optimization With Policy Feedback

2021 ◽  
pp. 1-11
Author(s):  
Yang Gu ◽  
Yuhu Cheng ◽  
C. L. Philip Chen ◽  
Xuesong Wang
Keyword(s):  
Policy Feedback ◽  
Policy Optimization

EVALUASI PROGRAM/PROYEK PENGEMBANGAN KAWASAN EKONOMI KHUSUS SEI MANGKE DALAM RANGKA PELAKSANAAN KEBIJAKAN MP3EI 2011-2025 DI KORIDOR EKONOMI SUMATERA

2018 ◽  
Vol 6 (1) ◽  
Author(s):  
Firrean Firrean
Keyword(s):  
Local Government ◽  
Evaluation Process ◽  
Spatial Arrangement ◽  
Policy Program ◽  
Evaluative Research ◽  
Policy Optimization ◽  
Economic Zones ◽  
North Sumatra

Special Economic Zones (SEZ) is a region with certain limits within the jurisdiction of Indonesia which is set to perform economic functions and obtain certain facilities. One SEZ developed in North Sumatra Province and included in the National Strategic Area (KSN) Medan - Binjai - Deli Serdang - Karo is SEZ Sei Mangke. SEZ Sei Mangke is defined in PP 29 of 2012 on 27 February 2012 and is the first KEK in Indonesia which was inaugurated its operation by President Joko Widodo on January 27, 2015. KSN Mebidangro itself is an area of priority spatial arrangement because it has a very important influence nationally against state sovereignty, defense and state security, economic, social, cultural, and / or environment, including areas designated as world heritage. This research is an evaluative research intended to find out the end of a policy program in order to determine recommendation of last policy by using CIPO model which includes four stages: (1) context, (2) input, (3) process, and (4) output. The research method used is case study by applying qualitative research that aims to make an accurate interpretation of the characteristics of the object under study. Findings on the evaluation context indicate that the program is generally running well, but some aspects of synergy and policy optimization as well as financing support from central and local government need to be improved. In the input evaluation, and evaluation process some aspects need to be improved because the findings show the weakness of some aspects is the result of lack of synergy and optimization of policy and support from local government. Interesting from the evaluation of ouput is that with some weaknesses in the input and process components, it turns out the evaluation findings ouput show Seek Mangke SEZ development can still run well. The recommendation of this research is to improve the quality of policy synergy / program of SEZ Seek development by improving several aspects that are categorized in each stage of evaluation

scholarly journals Air Learning: a deep reinforcement learning gym for autonomous aerial robot visual navigation

2021 ◽  
Author(s):  
Srivatsan Krishnan ◽  
Behzad Boroujerdian ◽  
William Fu ◽  
Aleksandra Faust ◽  
Vijay Janapa Reddi
Keyword(s):  
Reinforcement Learning ◽  
Embedded System ◽  
Broad Class ◽  
Visual Navigation ◽  
Raspberry Pi ◽  
Latency Distribution ◽  
Hardware In The Loop ◽  
Resource Constrained ◽  
Aerial Robot ◽  
Policy Optimization

AbstractWe introduce Air Learning, an open-source simulator, and a gym environment for deep reinforcement learning research on resource-constrained aerial robots. Equipped with domain randomization, Air Learning exposes a UAV agent to a diverse set of challenging scenarios. We seed the toolset with point-to-point obstacle avoidance tasks in three different environments and Deep Q Networks (DQN) and Proximal Policy Optimization (PPO) trainers. Air Learning assesses the policies’ performance under various quality-of-flight (QoF) metrics, such as the energy consumed, endurance, and the average trajectory length, on resource-constrained embedded platforms like a Raspberry Pi. We find that the trajectories on an embedded Ras-Pi are vastly different from those predicted on a high-end desktop system, resulting in up to $$40\%$$ 40 % longer trajectories in one of the environments. To understand the source of such discrepancies, we use Air Learning to artificially degrade high-end desktop performance to mimic what happens on a low-end embedded system. We then propose a mitigation technique that uses the hardware-in-the-loop to determine the latency distribution of running the policy on the target platform (onboard compute on aerial robot). A randomly sampled latency from the latency distribution is then added as an artificial delay within the training loop. Training the policy with artificial delays allows us to minimize the hardware gap (discrepancy in the flight time metric reduced from 37.73% to 0.5%). Thus, Air Learning with hardware-in-the-loop characterizes those differences and exposes how the onboard compute’s choice affects the aerial robot’s performance. We also conduct reliability studies to assess the effect of sensor failures on the learned policies. All put together, Air Learning enables a broad class of deep RL research on UAVs. The source code is available at: https://github.com/harvard-edge/AirLearning.

Coverage Path Planning with Proximal Policy Optimization in a Grid-based Environment

2020 ◽  
Author(s):  
Aleksandr Ianenko ◽  
Alexander Artamonov ◽  
Georgii Sarapulov ◽  
Alexey Safaraleev ◽  
Sergey Bogomolov ◽  
...  
Keyword(s):  
Path Planning ◽  
Coverage Path Planning ◽  
Policy Optimization ◽  
Grid Based
Sign in / Sign up

Export Citation Format

Share Document