Collaborative robots have been widely employed to assist humans in an increasing number of areas. Just as human-human collaboration, the trust in human-robot teams has a property of bidirectional. However, few studies have been conducted on both human-trusting-robot issue and robot-trusting-human issue in a unified framework for human-robot collaboration. The project addresses this challenge by developing a new systematic Bi-Trust framework to integrate humans? trust in robots and robots? trust in humans into the human-robot collaboration process. With the established Bi-Trust framework, a trust-level-based computational collaboration model is created to optimize and plan robot actions. The proposed approaches will reduce uncertain failures and improve the collaboration-quality of human-robot shared tasks.