At the risk of giving an overly commercial answer, I can sell you a system to do that. 
You said outdoors, that makes GPS/GNSS (Global Navigation Satellite System, a catch all for systems that use a mix of GPS, Glonass, Galileo, Beidou and other regional systems) by far the simplest option if it will work in your application. Off the shelf basic GNSS will give you 10 Hz update, ~1m accuracy (but better than that short term repeatability) and is cheap. High end systems will get you down to a couple of cm error and 100 Hz update rates but cost a lot more. There are a whole range of options in between.
If you only have partial GNSS coverage (e.g. some areas the view of the sky is obscured) a combined IMU / GNSS solution would be the best option. Generally this will cost a lot more than GNSS only.
(I could supply you with any of the above as off the shelf products targeted to the automotive market. So could quite a few other companies.)
If GNSS isn’t an option then optical or UWB are alternatives. Either on their own or combined with inertial.
Depending on latency requirements and number of possible obstructions optical may not be ideal in this situation but may be possible but I have seen some fairly impressive results from combining optical and inertial sensors. This has the advantage of not requiring the infrastructure that UWB requires.
The off the shelf decawave system doesn’t do what you need but the hardware can do this with the correct firmware.
We have a combined UWB/IMU kalman filtered system which will give 5-10cm accurate positions at speeds up to 200 km/h. Data is output at the tag with a latency of ~60 ms.
The system can also transition between GNSS and UWB as needed (see the second video here for an example New product announcement - Racelogic indoor positioning for automotive testing)
In terms of UWB coverage area we need a grid of anchors at around 30 m spacing. Up to 200 anchors are supported meaning a 30m x 3 km area could be covered. If needed we could probably push this to just under 250 anchors or just over 3.5 km. The anchors only require power, no network or other cable requirements. But if you can use GNSS for some of the area then that section doesn’t need UWB coverage so 5 km may be possible depending on the site. Spacing could be increased if you could get the appropriate licenses to bump the transmit power up but how practical that is is very region and use case specific.
The downside to the system is that we don’t support many tags simultaneously, currently a max of 5 but depending on the use case there may be some games you could play to increase that. And that we’ve spent a lot of time getting this right, we’re going to charge you a lot more than an off the shelf decawave module.
The final option would be a pure inertial system. However depending on your requirements this would probably be too expensive for you, pure inertial systems without some other reference to constrain the drift either lack the accuracy you need or cost silly money