It seems to me that you have a concept but no idea how to implement it. I assume this is for an academic project.
Voice recognition can be done on a PC or on the Arduino. For a real wheelchair this would need to be done on the wheelchair - the Arduino could do this using EasyVR or similar. As a proof of concept, I expect you'd find that part easier to implement on a PC.
Controlling drive motors is a separate issue but the challenges here are mainly on the electrical side. Do you have any experience of building/using electronic speed controls? If not, you need to look for a commercial implementation, which will probably be expensive by the time you're handling enough current to power a wheelchair.
Collision detection / avoidance is another area that can be complex to get working reliably in any sort of real-world environment.
Do you have any relevant knowledge and experience in any of these areas? If not, I recommend that you look for something less ambitious, or at least plan the project in stages so that if you don't complete the whole thing you do at least have something to show.
If you're planning to have a real usable product at the end, I suggest you need to rethink. This sounds like a safety critical project and not at all suitable for a novice, which you seem to be.