Dataset Information

An Embodied Agent Learning Affordances With Intrinsic Motivations and Solving Extrinsic Tasks With Attention and One-Step Planning.

ABSTRACT: We propose an architecture for the open-ended learning and control of embodied agents. The architecture learns action affordances and forward models based on intrinsic motivations and can later use the acquired knowledge to solve extrinsic tasks by decomposing them into sub-tasks, each solved with one-step planning. An affordance is here operationalized as the agent's estimate of the probability of success of an action performed on a given object. The focus of the work is on the overall architecture while single sensorimotor components are simplified. A key element of the architecture is the use of "active vision" that plays two functions, namely to focus on single objects and to factorize visual information into the object appearance and object position. These processes serve both the acquisition and use of object-related affordances, and the decomposition of extrinsic goals (tasks) into multiple sub-goals (sub-tasks). The architecture gives novel contributions on three problems: (a) the learning of affordances based on intrinsic motivations; (b) the use of active vision to decompose complex extrinsic tasks; (c) the possible role of affordances within planning systems endowed with models of the world. The architecture is tested in a simulated stylized 2D scenario in which objects need to be moved or "manipulated" in order to accomplish new desired overall configurations of the objects (extrinsic goals). The results show the utility of using intrinsic motivations to support affordance learning; the utility of active vision to solve composite tasks; and the possible utility of affordances for solving utility-based planning problems.

SUBMITTER: Baldassarre G

PROVIDER: S-EPMC6676802 | biostudies-literature | 2019

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

An Embodied Agent Learning Affordances With Intrinsic Motivations and Solving Extrinsic Tasks With Attention and One-Step Planning.

Baldassarre Gianluca G Lord William W Granato Giovanni G Santucci Vieri Giuliano VG

Frontiers in neurorobotics 20190726

We propose an architecture for the open-ended learning and control of embodied agents. The architecture learns action affordances and forward models based on intrinsic motivations and can later use the acquired knowledge to solve extrinsic tasks by decomposing them into sub-tasks, each solved with one-step planning. An affordance is here operationalized as the agent's estimate of the probability of success of an action performed on a given object. The focus of the work is on the overall architec ...[more]

PMID: 31402859

Similar Datasets

Project description:Overhunting typically increases during and after armed conflicts, and may lead to regional-scale defaunation. The mitigation of hunting impacts is complex because, among other reasons, several intrinsic and extrinsic motivations underpin the elevated deployment of hunting practices. Here we present the first study focusing on these motivations in a post-war zone. Following persistently heavy hunting pressure during the 27-year Angolan civil war, the offtake of small to medium-bodied species has increased recently as a result of large mammal depletion. However, prey choice associated with different motivations varied in terms of species trophic level and body size. While most residents hunted large-bodied species to maximize revenues from wildlife trade, many low-trophic level smaller species were harvested to meet local subsistence demands because they were more palatable and could be captured using artisanal traps near hunters' households. Mainly low-trophic level species were killed in retaliation for crop-raiding or livestock depredation. Considering all game species sampled in this study, 96% were captured to attend two or more motivations. In addition, hunting associated with different motivations was partitioned in terms of age and gender, with prey acquisition for the wildlife trade primarily carried out by adult men, while hunting to meet local subsistence needs and inhibit human-wildlife conflicts were carried out by adult men and women, children and even the elderly. In natural savannah areas lacking fish as a source of protein, a higher number of species was selected to supply both the meat trade and subsistence, while more species in forest areas were targeted for trade in animal body parts and conflict retaliation. Finally, local commerce in bushmeat and other body parts accrued higher domestic revenues compared to any alternative sources of direct and indirect income. However, these financial benefits were at best modest, largely unsustainable in terms of prey population collapses, and generated high long-term costs for the local to regional scale economy and native biodiversity.

Dataset Information

An Embodied Agent Learning Affordances With Intrinsic Motivations and Solving Extrinsic Tasks With Attention and One-Step Planning.

Publications

An Embodied Agent Learning Affordances With Intrinsic Motivations and Solving Extrinsic Tasks With Attention and One-Step Planning.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets