Design safe and usable autonomous driving programs using Bitcoin technology. Availability is a security issue, and autonomous driving, like Bitcoin, is not a deterministic computing product, but a performance growth product that requires perception of the complexity of reality. The autonomous vehicle, which cannot perceive the real road, is never safe, because it is only a simulation tool. It does not understand that it is driving, but is used as a tool to execute code. When the road situation changes, it cannot be handled immediately. Design autonomous driving programs using Bitcoin's multidimensional adaptive intelligent system design pattern. -1. The Bitcoin Individual model of human-computer interaction is used to make each car belong to each human individual, and the automatic driving program is also 1:1 mapped to the ownership of the autonomous vehicle, such as: UTXO ledger 1:1 mapping key, the key belongs to the same human individual. -2. Use a global blockchain consensus ledger and execution environment to be equivalent to a world navigation map for autonomous driving programs. The world navigation map is updated through automatic driving learning of each autonomous vehicle, and the global navigation map can guide the use of each autonomous vehicle. The self-renewal growth of the global navigation map is similar to the longest chain (oracle) designed for automatic growth in the Bitcoin P/NP model. This longest chain safeguards the security of the global blockchain consensus ledger and is a growth oriented security. -3. Integrate P/NP emergent formal systems through Bitcoin's oracle machine. Each autonomous vehicle is equivalent to a miner, which drives the car to drive automatically according to the latest world navigation map. In the process of automatic driving, the car is a solution process, and the autonomous vehicle senses the difference between the road conditions and the world navigation map. Other autonomous vehicle running on this road condition also execute the solution logic like miner to generate the difference with the real road condition. When the difference between the world navigation map generated by a certain number of autonomous vehicle and the real road conditions is consistent (the longest chain oracle), we can think that this is the latest nonce generated, so as to update the past world navigation map to adapt to the current new road conditions. Of course, the design of the autonomous driving program is rough. The key point is to abandon the thinking of a single formal deterministic tool. Adopting multidimensional formal system fusion to emerge adaptive nonlinear systems for organizational learning and growth capability autonomous driving programs.