ish D* algorithm more staki than A* algorithm ??

[orhneeorh]

but tiagong D* lite even better ???

 

[davidktw]

My limited understanding is the difference in time between the 2 algorithms at which you arrive at an optimal path.
It's not necessary the best but given a sub-optimal scenario, D* will arrive at an optimal path much faster than A* having D* doesn't discard the calculated path. It is builds on its past and hence incremental. A* on the other hand always need to start from scratch, which I assume can get a more optimum path, but at the expense of speed.
Dijkstra shortest path should be the slowest but can produce the best path.

Hence there is no best here. It depends on your application. All heuristic path searching algorithms may not result in the best path. As long as you are satisfied with the optimal path, then it will work for you.
It largely depends on your COST. Incremental heuristic will be favourable if the cost of calculation is significant vs the cost to move. If fuel cost is of absolute to you and you absolutely need the shortest path, then even A* might not be sufficient.

Analogy: There is no best price to buy in a stock, because time is of essence and opportunity cost matters. You can only depend on limited information to make the best sub-optimal decision available to you.

That is why D* Lite is good for robotics path planning since the robot will not likely get the full picture of the map and in cases where obstacles are changing all the time, you need an incremental approach to solve a dynamic situation. Robots nowadays uses SLAM to be informed on the most current surrounding and D* Lite fit in the scenario perfectly.