Following and interpreting narrated guided tours

Hemachandra, Sachithra; Kollar, Thomas; Roy, Nicholas; Teller, Seth

doi:10.1109/icra.2011.5980209

Cited by 44 publications

(42 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Gockley et al [26] used a similar approach, with a few modifications, including using a Brownian motion model for the tracking component. This approach was further extended by Hemachandra [31], which improved the person-following component by proposing a navigation approach that accounts for personal space, while avoiding obstacles.…”

Section: Using Only Laser Scannersmentioning

confidence: 99%

“…Nonetheless, variable environment lighting, backgrounds and appearances of people are factors which can be difficult to control for and can mislead vision-reliant tracking systems. [62] laser indoor moving local minima SJPDAF none Topp et al [69] laser indoor leg shapes (heuristic) SJPDAF ∼ potential field Gockley et al [26] laser indoor leg shapes (heuristic) NN direct & path Hemachandra et al [31] laser indoor leg shapes (heuristic) NN ∼ potential field Arras et al [2] laser indoor leg shapes (ML) MHT none Lu et al [44] laser indoor leg shapes (ML) NN none Munaro et al [52] RGB-D indoor height & HOG GNN none Gritti et al [28] RGB-D indoor leg shapes (ML) NN & PDAF none Cosgun et al [16] laser & RGB-D indoor leg shapes (heuristic) NN ∼ dynamic window Kobilarov et al [37] laser & omni-camera outdoor person shapes (heuristic) PDAF direct & path Navarro-Serment et al [53] laser outdoor person shapes (heuristic) NN none…”

Section: Using Other Sensor Modalitiesmentioning

confidence: 99%

See 1 more Smart Citation

Person tracking and following with 2D laser scanners

Leigh

Pineau

Olmedo

et al. 2015

2015 IEEE International Conference on Robotics and Automation (ICRA)

162

127

View full text Add to dashboard Cite

ACKNOWLEDGEMENTSThis thesis would not have been possible without the help of a great many people who provided me with inspiration, direction and encouragement.Foremost, I would like to thank my supervisor, Joelle Pineau. You've been the ideal mixture of supportive, easy-going and motivating during my master's. You were at once hands-off, letting me pursue the research ideas I was most interested in, and engaged, constantly providing me with valuable feedback on whatever we were working on. You listened to my ideas from day one so I would grow and learn to think for myself rather than follow orders. You were patient when research went slow and were quick to offer praise when things went well. I could not have asked for more from a supervisor. In this thesis we propose a new algorithm for robust detection, tracking and following from laser scanners mounted at leg height. We develop a set of quantitative benchmarks based on the CLEAR MOT metrics which demonstrate that the approach is more effective than currently available alternatives in a variety of environments, both indoor and outdoor, and on different robot platforms (the intelligent power wheelchair and a Clearpath Husky). To the best of our knowledge, these are the first benchmarks which evaluate person tracking performance from a mobile robot with laser scanners at leg height. We further integrate our novel person tracking system with a person following control system to demonstrate its usefulness in this application. The method has been implemented in the Robot Operating System (ROS) framework and will be publicly released as a ROS package. Dans cette thèse nous proposons un nouvel algorithme pour la détection, la localisation et le suivià partir de capteurs lasers situésà hauteur de la jambe.Nous développons un ensemble de mesures de référence quantitatives basées sur les métriques CLEAR MOT, qui démontrent que notre approche est plus efficace que les alternatives actuelles, ce dans une variété d'environnements,à la fois intérieurs et extérieurs, et sur différentes plateformes de robots (le fauteuil roulantélectrique intelligent et un Clearpath Husky). Au mieux de notre connaissance, nous sommes les premiersàévaluer la performance du suivi de personnes depuis un robot mobilé equipé de capteurs lasersà hauteur de la jambe. De plus, pour en démontrer l'utilité, nous intégrons notre nouveau système de localisation de personnes avec un système de contrôle automatique permettant au robot de se déplacer en suivant une personne.

show abstract

Section: Using Only Laser Scannersmentioning

confidence: 99%

Section: Using Other Sensor Modalitiesmentioning

confidence: 99%

Person tracking and following with 2D laser scanners

Leigh

Pineau

Olmedo

et al. 2015

2015 IEEE International Conference on Robotics and Automation (ICRA)

162

127

View full text Add to dashboard Cite

show abstract

“…Several researchers have augmented lower-level metric maps with higher-level topological and/or semantic information [7,8,9,10,11]. Zender et al [9] describe a framework for office environments in which the semantic layer models room categories and their relationship with the labels of objects within rooms.…”

Section: Related Workmentioning

confidence: 99%

Learning Semantic Maps from Natural Language Descriptions

Walter¹,

Hemachandra²,

Homberg³

et al. 2013

Robotics: Science and Systems IX

Self Cite

View full text Add to dashboard Cite

Abstract-This paper proposes an algorithm that enables robots to efficiently learn human-centric models of their environment from natural language descriptions. Typical semantic mapping approaches augment metric maps with higher-level properties of the robot's surroundings (e.g., place type, object locations), but do not use this information to improve the metric map. The novelty of our algorithm lies in fusing high-level knowledge, conveyed by speech, with metric information from the robot's low-level sensor streams. Our method jointly estimates a hybrid metric, topological, and semantic representation of the environment. This semantic graph provides a common framework in which we integrate concepts from natural language descriptions (e.g., labels and spatial relations) with metric observations from low-level sensors. Our algorithm efficiently maintains a factored distribution over semantic graphs based upon the stream of natural language and low-level sensor information. We evaluate the algorithm's performance and demonstrate that the incorporation of information from natural language increases the metric, topological and semantic accuracy of the recovered environment model.

show abstract

“…Unlike the SLAM problem, semantic mapping [23,51,19,15,36] is primarily interested in learning higher-level properties of the robot's environment. These properties include spatial attributes (like metric mapping) as well as concepts such each room's type (e.g., "hallway," or "kitchen"), their colloquial names (e.g., "Carrie's office"), or the objects that they contain.…”

Section: Related Workmentioning

confidence: 99%

A framework for learning semantic maps from grounded natural language descriptions

Walter

Hemachandra

Homberg

et al. 2014

The International Journal of Robotics Research

Self Cite

View full text Add to dashboard Cite

This paper describes a framework that enables robots to efficiently learn human-centric models of their environment from natural language descriptions. Typical semantic mapping approaches are limited to augmenting metric maps with higherlevel properties of the robot's surroundings (e.g., place type, object locations) that can be inferred from the robot's sensor data, but do not use this information to improve the metric map. The novelty of our algorithm lies in fusing high-level knowledge that people can uniquely provide through speech with metric information from the robot's low-level sensor streams. Our method jointly estimates a hybrid metric, topological, and semantic representation of the environment. This semantic graph provides a common framework in which we integrate information that the user communicates (e.g., labels and spatial relations) with metric observations from low-level sensors. Our algorithm efficiently maintains a factored distribution over semantic graphs based upon the stream of natural language and low-level sensor information. We detail the means by which the framework incorporates knowledge conveyed by the user's descriptions, including the ability to reason over expressions that reference yet unknown regions in the environment. We evaluate the algorithm's ability to learn human-centric maps of several different environments and analyze the knowledge inferred from language and the utility of the learned maps. The results demonstrate that the incorporation of information from free-form descriptions increases the metric, topological and semantic accuracy of the recovered environment model.

show abstract

Following and interpreting narrated guided tours

Cited by 44 publications

References 15 publications

Person tracking and following with 2D laser scanners

Person tracking and following with 2D laser scanners

Learning Semantic Maps from Natural Language Descriptions

A framework for learning semantic maps from grounded natural language descriptions

Contact Info

Product

Resources

About