In this paper, we first developed an artificial intelligence (AI)-edge emotion recognition platform using multimodal physiological signals wearable sensors: Electroencephalogram (EEG), electrocardiogram (ECG), and photoplethysmogram (PPG) sensors. Convolution neural network (CNN) was implemented here for 3 emotions (happy, angry, and sadness) classification. The EEG-based system with inputs from a short-time Fourier transform (STFT) pre-processing achieved a subject-independentaverageaccuracyof76.94%.TheECG/PPGbased system with features vector achieved an average subject-dependent accuracy of 76.8%. The proposed system was then integrated using the RISC-V processor and FPGA platforms to implement real-time monitoring and classification on edge. A 3to-1 Bluetooth network was also deployed here to transmit all physiological signals for the objective of low power.