Information Gain-based Exploration Using Rao-Blackwellized Particle Filters

By Cyrill Stachniss, Giorgio Grisetti, Wolfram Burgard. Robotics: Science and Systems Conference, 2005

605 citations recorded by the source corpus.

This paper presents an integrated approach to exploration, mapping, and localization. Our algorithm uses a highly efficient Rao-Blackwellized particle filter to represent the posterior about maps and poses. It applies a decision-theoretic framework which simultaneously considers the uncertainty in the map and in the pose of the vehicle to evaluate potential actions. Thereby, it trades off the cost of executing an action with the expected information gain and takes into account possible sensor measurements gathered along the path taken by the robot. We furthermore describe how to utilize the properties of the Rao-Blackwellization to efficiently compute the expected information gain. We present experimental results obtained in the real world and in simulation to demonstrate the effectiveness of our approach.

Subfields: Aerospace Engineering, Artificial Intelligence, Electrical and Electronic Engineering

View Paper: https://www.semanticscholar.org/paper/ee5afbd3737423115d159c2058f00279949e17d0

How researchers see this paper

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

In Aerospace Engineering

In Artificial Intelligence

In Electrical and Electronic Engineering

Citations in the corpus

Trajectory Optimization using Reinforcement Learning for Map Exploration (2008, result): “This idea has been pursued with some success (Sim et al. 2004 Stachniss et al. 2005), but even a one-step optimization is slow to compute.”
A Survey on Active Simultaneous Localization and Mapping: State of the Art and New Frontiers (2022, background): “[38] showed that combining distance and information-based functions results in better exploration strategies, and this has since been a common approach [128].”
Learning to soar: Resource-constrained exploration in reinforcement learning (2015, background): “…Hollinger et al. (2013) have exploited the submodularity property of information gain to select the most informative set of sensing locations, while Stachniss et al. (2005), Amigoni and Caglioti (2010) and Levine et al. (2010) also consider trade-offs between the information gathered and the cost…”
Active Policy Learning for Robot Planning and Exploration under Uncertainty (2007, background): “Our paper presents richer criteria for active learning as well suitable optimization objectives.”
An Exploration of Embodied Visual Exploration (2020, positive): “Reconstruction-based exploration approaches also relate to classical information-gain based exploration, where the goal is to explore in order to reduce the uncertainty in the agent’s model of the world (Cassandra et al. 1996; Sun et al. 2011; Stachniss et al. 2005).”
A Bayesian exploration-exploitation approach for optimal online sensing and planning with a visually guided mobile robot (2009, background): “Online path planning is essential for proper simultaneous localization and mapping (SLAM) [Sim and Roy (2005); Stachniss et al (2005)].”
Learning to Explore using Active Neural SLAM (2020, background): “, 2009; Kollar and Roy, 2008), and or seek to find a sequence of actions that minimizes uncertainty of maps (Stachniss et al., 2005; Carlone et al., 2014).”
Active Policy Learning for Robot Planning and Exploration under Uncertainty (2008, background): “For instance, full observability is assumed in [9, 7], known robot location is assumed in [10], myopic planning is adopted in [8], and discretization of the state and/or actions spaces appears in [11, 12, 7].”
Bayesian Learning for Safe High-Speed Navigation in Unknown Environments (2015, background): “Some exploration work has balanced the objectives of information gain, navigation cost and localization quality from a utility or decision-theory point of view [16, 26].”
Past, Present, and Future of Simultaneous Localization and Mapping: Toward the Robust-Perception Age (2016, neutral): “theoretic function, as the information gain, could be used to rank the different actions [23], [233].”
Learning Awareness Models (2018, methodology): “A sea of other methods have been proposed for exploration (MacKay, 1992; Ghavamzadeh et al., 2015; Asmuth et al., 2009; Gal, 2016; Stachniss et al., 2005; Plappert et al., 2017; Fu et al., 2017).”
Optimal Value of Information in Graphical Models (2009, background): “In the robotics literature, Stachniss, Grisetti, and Burgard (2005), Sim and Roy (2005) and Kollar and Roy (2008) have presented approaches to information gathering in the context of Simultaneous Localization and Mapping (SLAM).”
Seeing the Un-Scene: Learning Amodal Semantic Maps for Room Navigation (2020, methodology): “[8] and [35] use Rao-Blackwellized Particle Filters to choose the set of actions that maximize the information gain and minimize the uncertainty of the predicted maps.”
Imitation learning for natural language direction following through unknown environments (2013, methodology): “Our work is related to exploration strategies used for Simultaneous Localization and Mapping [14], as we must trade off the cost of executing actions with the cost of gathering information.”