FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2562651064> ?p ?o ?g. }

• W2562651064 abstract "Perception is the process by which sensor inputs are selected, organized and interpreted into the agent's knowledge of the world. The key challenge of perception is partial observability, which causes uncertainty in the robot's internal knowledge. For a mobile robot, robust perception can be achieved through probabilistic state estimation and active sensing. State estimation infers the state of the world from noisy percepts, while active sensing helps the robot to control its sensors to collect useful information. While progress has been made in recent years to improve these techniques, deploying them on mobile robots means that they have to face the complexity of the real world environments and have to be subjected to the constraints of limited real-time resources. This thesis is concerned with adapting robust perception techniques to real-time problems. The proposed solutions allocate resources efficiently according to their impact on uncertainty. Real-time environment poses computational constraints on state estimation algorithms. In particular, an important class of state estimation algorithm called particle filter requires a minimum amount of resource. It is necessary to enhance particle filters so that they can still perform robustly even when the minimum resource is not available. The proposed solution spreads the processing over multiple observations, and is capable of focusing computation on observations that have more impact on estimation. In addition, this thesis also tackles the problem of real-time state estimation in environments with high interactivity between the robot, its tracked targets and the environment. This estimation problem is solved by using Rao-Blackwellised particle filters. Noting its inefficiencies, an approximation is implemented which is shown to work in real-time with comparable accuracy. Using the information from state estimation, active sensing provides robust perception through sensor control. With limited sensing resource, the sensors can only deliver limited information, therefore, it is important that their usage is optimized. In this thesis our goal is to find an optimal sensing strategy that allocate sensing resources efficiently. Such strategy should supply relevant information to the robot according to different situations. A reinforcement learning solution to this problem is developed by using augmented representations that encode uncertainty in the state space. The empirical results suggest that this is a highly viable approach for learning effective sensing strategies." @default.
• W2562651064 created "2017-01-06" @default.
• W2562651064 creator A5014707300 @default.
• W2562651064 creator A5028722695 @default.
• W2562651064 date "2004-01-01" @default.
• W2562651064 modified "2023-09-24" @default.
• W2562651064 title "Robust real-time perception for mobile robots" @default.
• W2562651064 hasPublicationYear "2004" @default.
• W2562651064 type Work @default.
• W2562651064 sameAs 2562651064 @default.
• W2562651064 citedByCount "1" @default.
• W2562651064 crossrefType "journal-article" @default.
• W2562651064 hasAuthorship W2562651064A5014707300 @default.
• W2562651064 hasAuthorship W2562651064A5028722695 @default.
• W2562651064 hasConcept C154945302 @default.
• W2562651064 hasConcept C157286648 @default.
• W2562651064 hasConcept C19966478 @default.
• W2562651064 hasConcept C2776010242 @default.
• W2562651064 hasConcept C28826006 @default.
• W2562651064 hasConcept C33923547 @default.
• W2562651064 hasConcept C36299963 @default.
• W2562651064 hasConcept C41008148 @default.
• W2562651064 hasConcept C49937458 @default.
• W2562651064 hasConcept C52421305 @default.
• W2562651064 hasConcept C79403827 @default.
• W2562651064 hasConcept C90509273 @default.
• W2562651064 hasConceptScore W2562651064C154945302 @default.
• W2562651064 hasConceptScore W2562651064C157286648 @default.
• W2562651064 hasConceptScore W2562651064C19966478 @default.
• W2562651064 hasConceptScore W2562651064C2776010242 @default.
• W2562651064 hasConceptScore W2562651064C28826006 @default.
• W2562651064 hasConceptScore W2562651064C33923547 @default.
• W2562651064 hasConceptScore W2562651064C36299963 @default.
• W2562651064 hasConceptScore W2562651064C41008148 @default.
• W2562651064 hasConceptScore W2562651064C49937458 @default.
• W2562651064 hasConceptScore W2562651064C52421305 @default.
• W2562651064 hasConceptScore W2562651064C79403827 @default.
• W2562651064 hasConceptScore W2562651064C90509273 @default.
• W2562651064 hasLocation W25626510641 @default.
• W2562651064 hasOpenAccess W2562651064 @default.
• W2562651064 hasPrimaryLocation W25626510641 @default.
• W2562651064 hasRelatedWork W1557398617 @default.
• W2562651064 hasRelatedWork W1607872737 @default.
• W2562651064 hasRelatedWork W183933562 @default.
• W2562651064 hasRelatedWork W2027416654 @default.
• W2562651064 hasRelatedWork W2155847035 @default.
• W2562651064 hasRelatedWork W2183499408 @default.
• W2562651064 hasRelatedWork W2184942639 @default.
• W2562651064 hasRelatedWork W269697800 @default.
• W2562651064 hasRelatedWork W2767801795 @default.
• W2562651064 hasRelatedWork W2800532881 @default.
• W2562651064 hasRelatedWork W2892186189 @default.
• W2562651064 hasRelatedWork W2962926894 @default.
• W2562651064 hasRelatedWork W2991435565 @default.
• W2562651064 hasRelatedWork W3047400633 @default.
• W2562651064 hasRelatedWork W3080622745 @default.
• W2562651064 hasRelatedWork W3173439324 @default.
• W2562651064 hasRelatedWork W3186806860 @default.
• W2562651064 hasRelatedWork W756091146 @default.
• W2562651064 hasRelatedWork W939205321 @default.
• W2562651064 hasRelatedWork W1996982349 @default.
• W2562651064 isParatext "false" @default.
• W2562651064 isRetracted "false" @default.
• W2562651064 magId "2562651064" @default.
• W2562651064 workType "article" @default.