In the face of the challenges posed by the aging population with a declining birth rate in terms of life care issues, this paper proposes an intelligent smart home based on multi-heterogeneous elderly-care robots and their safety care methods. The aim is to address the shortage of professional caregivers and meet the safety assistance and continuous behavioral care needs of elderly individuals with reduced capabilities. The system incorporates seven heterogeneous elderly-care robots and on-board multiple sensor systems, covering various care tasks from delivery, getting up, transferring, walking, excretion, to indoor rehabilitation training. To ensure the efficiency and timeliness of task allocation in a multi-robot system during execution, a dynamic parallel auction execution algorithm is proposed, effectively handling the allocation of multi-priority tasks and emergency tasks. This algorithm can effectively handle the allocation of multi-priority tasks and emergency tasks, ensuring efficient and timely task execution. Experiments conducted in multi-task care scenarios and real-life home environments have verified the feasibility and adaptability of the proposed system architecture and methods in certain care scenarios, providing an effective solution to improve the quality of life for elderly individuals and alleviate the pressure on caregivers.