Noname manuscript No. (will be inserted by the editor) TEXPLORE: Real-Time Sample-Efficient Reinforcement Learning for Robots

Citations in the corpus

Deep Q-learning From Demonstrations (2017, neutral): “The learning framework that we have presented in this paper is one that is very common in real world problems such as controlling data centers, autonomous vehicles (Hester and Stone 2013), or recommendation systems (Shani, Heckerman, and Brafman 2005).”
Challenges of Real-World Reinforcement Learning (2019, background): “A common approach is to learn ensembles of transition models and use various sampling strategies from those models to drive exploration and improve sample efﬁciency (Hester & Stone, 2013; Chua et al., 2018; Buckman et al., 2018).”
Reinforcement Learning in non-stationary environments: An intrinsically motivated stress based memory retrieval performance (SBMRP) model (2014, background): “Sample efficiency problem as described by Hester [3] is very important to enable feasibility in physical robot applications where reduced sample training can save huge amount of learning time (thousands of physical robot’s actions is slow) in physical world.”
The Open-Source TEXPLORE Code Release for Reinforcement Learning on Robots (2013, methodology): “The full details of this architecture are described in [7, 9].”
A Survey of State-Action Representations for Autonomous Driving (2018, background): “For instance, Hester and Stone (2006); Shalev-Shwartz et al. (2016a) only consider the choice of continuous acceleration while assuming that a lane keeping lateral controller is available and independently sets the steering angle.”
Value-Aware Loss Function for Model-based Reinforcement Learning (2017, methodology): “There are many variations to each step of this generic algorithm such as how to collect new data points (cf. [Hester and Stone, 2013]) or what Planner should we use from all possible value-based, policy search, etc. algorithms.”
An empirical investigation of the challenges of real-world reinforcement learning (2020, background): “Another common approach is to learn ensembles of transition models and use various sampling strategies from those models to drive exploration and improve sample efficiency [Hester and Stone 2013; Chua et al. 2018; Buckman et al. 2018].”
A Neuroevolution Approach to General Atari Game Playing (2014, background): “One potentially fruitful avenue would be to apply intrinsically motivated exploration algorithms such as novelty search [26] or targeted exploration [19] to these games so that the intrinsic reward signal could act as a stand-in until the first real reward is encountered.”
Efficient PAC-Optimal Exploration in Concurrent, Continuous State MDPs with Delayed Updates (2016, background): “While the experimental results of Hester et al. are very encouraging, their algorithm does not come with any performance guarantees, nor does it handle concurrent exploration in multiple MDPs (though extending it to the multiple MDP case appears straightforward).”
Reset-free Trial-and-Error Learning for Robot Damage Recovery (2016, methodology): “In particular, TEXPLORE is a non-episodic model-based RL algorithm that is based on learning the transition dynamics of the robot, which are then used by an MCTS-based planner to select the most promising action in the current time step or episode [23].”
Exploration from Demonstration for Interactive Reinforcement Learning (2016, methodology): “There are several value-based like UCB [6] and others [24, 7, 11] and model-based methods [35, 29, 18, 19], studied in the context of multi-arm bandits, that perform eﬀective exploration by maintaining statistics about changes in the value function and the number of times state-action pairs have…”