They now have a txt message system that should be relied on:
1. It has translation built in.
2. It has pre-canned responses
In practice, with #2, the most common pre-canned response you give is "follow navigation" and my drivers tended to just respond with the pre-canned "ok". Drivers don't call me -- their app even tells them I'm not using a Chinese language app.
Never had any trouble.
It is not perfect, but this method allows me to move around easily with didi without speaking chinese. I had no urban mapping problems, but I do test whatever address I want to go in baidu maps first, or any other chinese map app. That way, in case of problems, I can just show him the location or use whatever voice guidance, but I rarely needed this.