Numerical Experiments

In this page, we provide with the full example of visiting cities.
We describe how we obtain the benchmark.

Configuration

We disign virtual ISS by shifting them of 1, 2 or 3 days.

from spot.rl.environments.satellite import SatelliteEnvironment

# ISS station
tle_string = ["1 25544U 98067A   17314.40254821  .00006490  00000-0  10525-3 0 9997",
              "2 25544  51.6429  29.4166 0004559 104.3372 354.3186 15.54111847 84492"]

# compute satellites' trajectories
satellites = []
for satellite_id in [1, 2, 3]:
    satellite = SatelliteEnvironment(satellite_id)
    satellite.init_from_tle(tle_string)
    satellites.append(satellite)

Observation requests

We sample random cities on earth.

num_requests = 50
requests = np.genfromtxt(f"data/requests_{num_requests}.txt")
priorities = np.ones(len(requests))

Solvers

We use the classical NetworkXSolver for illustration.

from spot.mis.networkx import NetworkXSolver
delta_t = 10

# prepare solver
solver_params={
    "prefix": f"nx_{delta_t}_{len(requests)}_{len(satellites)}",
    "delta_t": delta_t,
}

solver = NetworkXSolver(solver_params)

# run solver
solver.run(dtos, requests, priorities, satellites)

Metrics

When the solver runs, it collects some metrics.

Global satisfaction rate

The satisfaction rate is the number of completed requests in percentage.
It can be splited by priorities.

Time to solution

Time to solution recovers elapsed time for the solver to extract the MIS.

Image quality

Image quality is distance to 0 pitch (second component of the satellite attitude).
The closest to 0 the better.

Workload balance between satellites

We can evaluate how well the plan is compared to the balancing evenly the workload between satellites.

Maneuver duration

The maneuver duration needs to be minimized as it consumes energy onboard and delays the opportunities.

Global plan results

Here we show an example of the globla plan.

"results": {
    "completion_rate":
    {
        "global_completion_rate": 6.0,
            "prioritized_completion_rates": {"1.0": 6.0}
    },
    "time_to_solution": 30.29348,
    "workload_balance": {"0": 4.0, "1": 0, "2": 2.0},
    "image_quality": NaN,
    "maneuver_duration": NaN,
    "overlap_complexity": 0,
    "global_plan": {"7": [0, 10], "16": [2, 2610], "23": [0, 30]}
}