Emotional Model Based on Computational Intelligence for Partner Robots

Kubota, Naoyuki; Wakisaka, Shiho

doi:10.1007/978-3-642-12604-8_5

Cited by 5 publications

(3 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Some use for example fuzzy logic rules to connect input to emotions (Ayesh, 2004). Another example we encountered is the previous emotional state (at t−1) influencing the current emotional state (Kubota and Wakisaka, 2010). An example is the Markovian transition model between emotions in (Ficocelli et al, 2015), with similar ideas in (Zhang and Liu, 2009).…”

Section: Methods Papersmentioning

confidence: 99%

“…Value (Bozinovski, 1982;Bozinovski et al, 1996) (Broekens et al, 2007a) (Schweighofer and Doya, 2003) (Hogewoning et al, 2007) (Shi et al, 2012) (Blanchard and Canamero, 2005) Reward (Moren and Balkenius, 2000;Balkenius and Morén, 1998) (Ahn and Picard, 2006) implementations use the detected emotional state of another person to influence the emotion of the agent/robot (Hoey et al, 2013) (Ficocelli et al, 2015). Hasson et al (2011) uses facial expression recognition systems to detect human emotion, while Kubota and Wakisaka (2010) uses human speech input. Note that if these agent emotions subsequently influence agent learning, then we come very close to learning from human emotional feedback (as briefly described in Section 2.3).…”

Section: Methods Papersmentioning

confidence: 99%

“…For example, both Tsankova (2002) and Hasson et al (2011) use a high frustration to switch between behaviour. Similarly, Kubota and Wakisaka (2010) use several emotions to switch between the weighting of different value functions. For example, happiness leads to exploration by selecting a value function derived from inverse recency.…”

Section: Action Selectionmentioning

confidence: 99%

See 2 more Smart Citations

Emotion in reinforcement learning agents and robots: a survey

2017

View full text Add to dashboard Cite

This article provides the first survey of computational models of emotion in reinforcement learning (RL) agents. The survey focuses on agent/robot emotions, and mostly ignores human user emotions. Emotions are recognized as functional in decision-making by influencing motivation and action selection. Therefore, computational emotion models are usually grounded in the agent's decision making architecture, of which RL is an important subclass. Studying emotions in RL-based agents is useful for three research fields. For machine learning (ML) researchers, emotion models may improve learning efficiency. For the interactive ML and human-robot interaction community, emotions can communicate state and enhance user investment. Lastly, it allows affective modelling researchers to investigate their emotion theories in a successful AI agent class. This survey provides background on emotion theory and RL. It systematically addresses (1) from what underlying dimensions (e.g. homeostasis, appraisal) emotions can be derived and how these can be modelled in RL-agents, (2) what types of emotions have been derived from these dimensions, and (3) how these emotions may either influence the learning efficiency of the agent or be useful as social signals. We also systematically compare evaluation criteria, and draw connections to important RL sub-domains like (intrinsic) motivation and model-based RL. In short, this survey provides both a practical overview for engineers wanting to implement emotions in their RL agents, and identifies challenges and directions for future emotion-RL research.

show abstract