Planning under Uncertainty for Robotic Tasks with Mixed Observability

By S. W. Ong, S. Png, David Hsu, Wee Sun Lee. 2010

299 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Reinforcement Learning, Robotics & Control

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Reinforcement Learning

In Robotics & Control

Citations in the corpus

Towards Continual Reinforcement Learning: A Review and Perspectives (2020, background): “Alternatively, we can consider Mixed Observability MDPs (MOMDPs) (Ong et al., 2010), which are a more general formulation than passive settings, where the agent itself can potentially impact the way that the task state evolves.”
Deliberation for autonomous robots: A survey (2017, methodology): “Such a partial observability approach is pursued with the MOMDP model [166, 8] which represents the states space as the cartesian product of a set of visible states and a set of hidden states.”
Potential-based reward shaping for finite horizon online POMDP planning (2016, methodology): “To speed up computation in each benchmark, we used the state-of-the-art equivalent MOMDP7 representation [15] for the POMDP model, as also done in the recent online POMDP planning literature (e.”
Real-time motion planning methods for autonomous on-road driving: State-of-the-art and future research directions (2015, background): “Unlike MDPs, which assume that the states are fully observable, partially observable Markov Decision Processes (POMDPs) assume that the state of a robot or a vehicle is not known (Ong et al., 2010).”
Integrated task and motion planning in belief space (2013, background): “Recent pointbased POMDP solvers (Kurniawati et al., 2008) can be leveraged to solve a variety of robotics problems (Hsiao et al., 2007; Ong et al., 2010; Kurniawati et al., 2011), but mobile manipulation problems involving many movable objects are beyond the scope of these methods.”
Hidden Parameter Markov Decision Processes: A Semiparametric Regression Approach for Discovering Latent Task Parametrizations (2013, background): “However, solving POMDPs in high-dimensional, continuous state spaces remains challenging, despite advances in POMDP planning [4, 12, 6], including situations with mixed-observability [13].”
Sequential Decision-Making under Non-stationary Environments via Sequential Change-point Detection (2014, methodology): “Recently this model has also been described in the context of MOMDPs [12, 3].”
Search and pursuit-evasion in mobile robotics (2011, background): “A state-of-the-art approach that involves sophisticated sampling of the belief space was proposed by Ong et al. (2010) to allow for mixed observability (e.g., when the locations of the searchers are completely known but the location of the target is unknown).”
DESPOT: Online POMDP Planning with Regularization (2013, background): “The results for SARSOP and AEMS2 are replicated from [14] and [17], respectively.”
Perception, Planning, Control, and Coordination for Autonomous Vehicles (2017, background): “POMDPs assume uncertainty in both the robot’s motion and in observation and account for this uncertainty in solving for optimal policies, where MOMDPs have extended this idea to situations in which some of the state variables are partially observable and others are full observable [215].”