Additional Constraints

[krdee1]

Additional constraints for obstacles/domain and communications network

• Make sure first test case doesn't always only have 2 agents
• Profile code to make faster for debugging on laptop
• Automatically generate obstacle spanning the floor up to a fixed altitude (minimum altitude constraint)
• Write single agent single obstacle avoidance unit test
• Implement obstacle avoidance
• Implement domain constraint with CBF (walls and ceiling constraint)
• Implement domain min/max altitude constraints (obstacle spanning floor, domain constraint itself for ceiling)
• Raise obstacles to the minimum altitude to avoid intersection with the minimum altitude obstacle (looks nicer)
• Ensure that communications links are not established when the line of sight between two agents is blocked by an obstacle
• Add the ability to run without generating any plots for unit testing
• Add communications spheres to agents (like collision spheres)
• Plot motion trails of agents
• Implement lesser neighbor algorithm in agent.run
• Implement connectivity constraint as opposite of collision avoidance constraint in communication radius spheres on lesser neighbor connections
• Plot constraint h functions w.r.t. time on separate plots (one for collision avoidance, one for domain, obstacles, lesser neighbor algorithm one might be challenging due to time evolution)
• Make a test case for vertical GA where the agent has to go over and under barriers that span across the domain to reach the objective
• Plot motion trails for agents in unique colors, matching performance plot colors
• Add a legend to the main 3D plot to label agents by trail color
• Copy trails to other spatial plots in the main figure
• Consider building collision geometry into agent initialization to simplify building test cases
• Add output time history data (matfile)
• Add the ability to generate a video from said matfile (would allow the sim to be run without plotting and then the plotting would be all executed at the end)

[krdee1]

Should it be possible to gain sensor performance for areas obscured by an obstacle?

[krdee1]

Should it be possible to gain sensor performance for areas obscured by an obstacle?

Obstacle-obstructed areas should also probably be excluded from the gradient ascent portion