System Implementation

 

Rovers capable of communicating on the network

The rovers are operating within a range that is perceivable by the cumulative sensing range of the entities in the system. A rover’s position can be determined by at least one other rover in the system. Each rover sweeps for a set time T to aid in the self-pose determination of neighboring (within the sweeper’s field of view) rover.

Say at a given instance the location of all the rovers is unknown
The system executes a predefined initialization process to determine the location of every rover. System – all entities capable communicating on the given network
The initialization starts at time t. For the purposes of illustration, let the current time be (t) =0

The first robot relative pose with respect to the ground station R0 is determined within the time interval, i.e., [0, T]
The R2 second rover’s relative pose with respect to the first rover R1 is determined within the time interval, i.e., [T, 2T]
The second robot performs its scheduled operation of sweeping for the time interval [2T, 3T]
At the end of T (i.e., completion of R0’s sweep action), rover 1 has been able to determine its pose relative to R0. During the time interval [0, T] as far as rover 1 is concerned, it has received a sweep signal, based on the pre-built (prior knowledge) look-up table, it has determined that this could only be possible if it was within the sweeping field of view of R0
Over the next sweeping time interval [T, 2T] (i.e., completion of R1’s sweep action), rover 2 has been able to determine its pose relative to R1. Based on the time keeping on the network, R2 has been able to cross-reference the signal received during the [T, 2T] could have been possible if it was within the field of view of R1.
 

Every time a sweeping laser is intercepted by a HTC VIVE tracker which is mounted on a rover, the very same  rover’s on-board computer gets an update relative pose. Thanks to our minimal operation of time-keeping on the network and prior knowledge (cloned into each entity in the system) of sweeping sequence of the rovers, the updated relative pose can be assigned a frame of reference. This information is then published in the network.

Status of pose estimate of each entity in the system.
Given that all the rovers are communicating on a given and same network, we can determine relevant information of rover Ri -{health, position, orientation, mean change in position (wrt to last known info.) over a period of time, [Relative pose of the rovers in the network]}
Similarly pose info about rover R2 can be determined.

As per our initial design choice to have a rover pose be determinable by at least one another rover operating within the same network, we will need an additional 360 degree rotational degree of freedom to track the rescue rover (leaves its prior neighbor). As it approaches the entrapped rover it can happen that the entrapped rover might loose track of rescuer for extended period of time which is dependent on the “safe rescue and approach plan” determined by the pair. During this process of rescuing if the rescuer has to approach from behind the sweeper’s field of view, we stand a chance of losing track of the rescuer as well (which is no longer able to intercept a sweep signal from any of the entities in the system).