With increasing life expectancy, the technological sub-field of elderly care is expected to rise in importance. This growing population has put a strain on current elderly care facilities and personnel. This research work is focused on designing and implementing a collaborative elderly care system (CECAS) using a companion robot and a wearable device that will enable the monitoring and companionship needed in assisted living communities and private homes. This system is meant to improve the automation in elderly care which will lead to a reduction in costs and an increase in availability. Two elderly care applications are used to demonstrate the capability of the proposed system, which includes: fall detection and response, and conversational assistance. In the fall detection and response application, a two-step method is proposed, which consists of a motion data based preliminary detection on the wearable device and a video-based final detection on the companion robot. To further reduce the operating costs and improve the system, parameters such as the image size and the neural network size are optimized through extensive experimental study. In the conversational assistance application, the robot is capable to conversate with the user to determine if external assistance is needed and acts as an extra measure of fall detection. The results of the experiments showcase that both the wearable device and the companion robot can be optimized for power-efficiency, speed, and retain fall detection performance.