Based on the smart home and facial expression recognition, this paper presents a cognitive emotional model for eldercare robot. By combining with Gabor filter, Local Binary Pattern algorithm(LBP) and k-Nearest Neighbor algorithm(KNN) are facial emotional features extracted and recognized. Meanwhile, facial emotional features put influence on robot’s emotion state, which is described in AVS emotion space. Then the optimization of smart home environment on the cognitive emotional model is specially analyzed using simulated annealing algorithm(SA). Finally, transition probability from any emotional state to a state of basic emotions is obtained based on the cognitive reappraisal strategy and Euclidean distance. The simulation and experiment have tested and verified the effective in reducing negative emotional state. Based on the smart home and facial expression recognition, this paper presents a cognitive emotional model for eldercare robot. By combining with Gabor filter, Local Binary Pattern algorithm (LBP) and k-Nearest Neighbor algorithm (KNN) are facial emotional features extracted and recognized . The emotional of putative on robot’s emotional state, which is described in AVS emotion space. Then the optimization of smart home environment on the cognitive emotional model is specifically analyzed using simulated annealing algorithm (SA). Finally, transition probability from any emotional state to a state of basic emotions is obtained based on the cognitive reappraisal strategy and Euclidean distance. The simulation and experiment have tested and verified the effective in reducing negative emotional state.