[TRACKER] Navigation test plan

[Scony]

In the upcoming weeks, I'm going to work on test cases for the navigation area. I'm going to use this issue as a progress tracker but also as a place where anyone can propose scenarios to implement.

Please note that many scenarios will be tested in 3D only since quite a lot of navigation code is common for 2D and 3D so we don't need to duplicate test cases in many instances.

Server tests

NavigationServer3D

General

• Server should be empty when initialized - checking if the internal state of server is correct after initialization
• Server should release valid RIDs

Agent

• Server should create a valid agent
• Server should not report dangling agent in ProcessInfo
• Server should not report removed agent in ProcessInfo
• Server should report agent with active map in ProcessInfo
• Server should be able to set/get basic agent params
• Server should be able to set/get all agent's params
• (...)

Map

• Server should create a valid map
• Server should yield default values in response to queries against empty map
• Server should report active map in ProcessInfo
• Server should not report active map in ProcessInfo
• Server should be able to set/get all map's params
• Server should report correct number of map's agents
• Server should report correct number of map's obstacles
• Server should report correct number of map's links
• Server should report correct number of map's regions
• Server should emit map_changed
• Server should emit navigation_debug_changed
• (...)

Link

• Server should create valid link
• Server should not report dangling link in ProcessInfo
• Server should report link with active map in ProcessInfo
• Server should be able to set/get all link's params
• (...)

Region

• Server should create valid region
• Server should not report dangling region in ProcessInfo
• Server should report region with active map in ProcessInfo
• Server should yield default values in response to queries against empty region
• (...)

Obstacle

• Server should create valid obstacle
• Server should be able to set/get all obstacle's params
• (...)

NavigationServer2D

• Server should be empty when initialized - checking if the internal state of server is correct after initialization
• (...)

Node tests

NavigationRegion3D

• New region should have valid RID
• Navigation region should bake navigation mesh correctly
• Navigation region should emit bake_finished
• Navigation region should emit navigation_mesh_changed
• (...)

NavigationRegion2D

• New region should have valid RID
• (...)

NavigationAgent3D

• New agent should have valid RID
• New agent should attach to default map
• (...)

NavigationAgent2D

• New agent should have valid RID
• New agent should attach to default map
• (...)

NavigationObstacle3D

• New obstacle should have valid RID
• New obstacle should attach to default map
• Obstacle estimates radius correctly
• Obstacle uses hardcoded radius if not estimated
• Obstacle changes navigation map correctly
• Obstacle is avoided by agent correctly
• (...)

NavigationObstacle2D

• New obstacle should have valid RID
• New obstacle should attach to default map
• Obstacle changes navigation map correctly
• Obstacle is avoided by agent correctly
• (...)

NavigationLink3D

• (...)

NavigationLink2D

• (...)

PRs

• Add initial navigation tests #73121
• Add initial NavigationObstacle and NavigationRegion tests #73700
• Add agent and map tests for 'NavigationServer3D' #78449
• Add more basic tests for NavigationServer3D #78480
• Add advanced NavigationServer3D tests #78667

This issue supplements #43440

[smix8]

Looking for navigation related benchmark ideas.

Benchmark ideas are written in format like this:
TYPE: [Groups it Belongs]: Name: Description

Examples:

• Exclusive:AStar3D: Create random map with 1000 inter connected points (probably using Delaunay3D), Benchmark solving it 1000 times from random 2 points
• Exclusive:Navigation: On a premade map, solve 1000 random paths between two points on the surface of the shapes.
• Agents: Benchmark 1000 moving agents in a map with local collision avoidance.

[Scony]

Looking for navigation related benchmark ideas.

Benchmark ideas are written in format like this: TYPE: [Groups it Belongs]: Name: Description

Examples:

• Exclusive:AStar3D: Create random map with 1000 inter connected points (probably using Delaunay3D), Benchmark solving it 1000 times from random 2 points
• Exclusive:Navigation: On a premade map, solve 1000 random paths between two points on the surface of the shapes.
• Agents: Benchmark 1000 moving agents in a map with local collision avoidance.

Some more examples to consider:

• Exclusive:Navigation[Agents]: Benchmark 10 maps with 100 moving agents each and with local collision avoidance.
• Exclusive:Navigation[Agents]: Benchmark 100 maps with 10 moving agents each and with local collision avoidance.
• Exclusive:Navigation[Agents]: Benchmark 1000 moving agents in a map with local collision avoidance and with 100 triangular, static obstacles spread across the map.
• Exclusive:Navigation[Agents/Queries]: Benchmark 1000 moving agents in a map with local collision avoidance while changing desired location (path recalculation) for 100 of them every second for 10 seconds.