The emerging autonomic computing technology has been hailed by world-wide researchers and professionals in academia and industry. Besides four key capabilities, well known as self-CHOP, we propose an additional self-regulating capability to explicitly emphasize the policy-driven self-manageability and dynamic policy derivation and enactment. Essentially, these five capabilities, coined as Self-CHROP, define an autonomic system along with other minor properties. Trusted computing targets guaranteed secured systems. Self-protection alone does not ensure the trustworthiness in autonomic systems. The new trend is to integrate both towards trusted autonomic computing systems. This paper presents a comprehensive survey of the autonomic and trusted computing paradigms and a preliminary conceptual architecture towards trustworthy autonomic grid computing.