I used National Panasonic video cameras in the 1980’s that employed ultrasonic distance measurement and motorised focus adjustment. These worked OK but I think the issue may be the shallow depth of field that thermal cameras can have. A shallow depth of field is very demanding on an auto focus system that does not analyse the actual image being captured.
The thermal cameras that I have with real auto focus use the image being collected and even those expensive cameras are not perfect at getting the focus right every time. Auto focus on a thermal camera is not easy, no matter how you approach it.