This feels more plausible for long-term success than Waymo for a number of reasons:
* Construction sites are smaller in footprint
* They’re more easily covered in sensor networks to support autonomous operations
* They’re typically controlled-access environments, which reduces potential interruptions to automated routines
* Since they have higher risk profiles than public infrastructure, the expectation is that workers will be more aware of autonomous operations and any deviations before they cause serious harm
Honestly, I’d wager closed-site autonomy takes off before anyone nails national or global self-driving on existing infrastructure.
The flip side is that they work with a lot more degrees of freedom (a car has just steering and speed to adjust), very diverse machinery and a ton more exceptions that need to be handled frequently
True, but they have more freedom to attack low-hanging fruits and build capabilities in order of ease. With a self-driving car, there’s not a whole lot of iteration allowed - you have your capital to do R&D, but eventually investors expect that car to drive on public roads without incident or support, something nobody can do at the moment.
For construction sites, on the other hand, you could (more) easily automate things like dump trucks or material transports using preprogrammed routes and manual triggers when something is done or needed. Then you can iterate on those systems to add more capabilities, more automations, more integrations with other equipment.
Kind of like how tractors have iteratively improved over time because they benefit from a lot of the same limitations as construction sites. At least that’s my thinking on it.
There are btw modern tractors that can drive autonomously on the fields. You configure the GPS coordinates of the field and the tractor can do various routes and patterns through the field on it's own. You're just not allowed to leave the wheel alone with current laws.
Definitely they can keep slicing the problem and deliver viable solutions within the boundary specified.
I just don’t think that a single solution will cover the full suite of construction site machinery. Scaling will be tough. But as you said they will be printing cash in the meantime.
* Construction sites are smaller in footprint
* They’re more easily covered in sensor networks to support autonomous operations
* They’re typically controlled-access environments, which reduces potential interruptions to automated routines
* Since they have higher risk profiles than public infrastructure, the expectation is that workers will be more aware of autonomous operations and any deviations before they cause serious harm
Honestly, I’d wager closed-site autonomy takes off before anyone nails national or global self-driving on existing infrastructure.