Types of mission testing ======================================================= (a) A human operator launches a mission, observes the activity, and after some period of time decides the mission has provided enough information. The user halts the mission and rates the result. Launch (Human) Halt (Human) Assess (Human) (b) A human operator launches, auto-completes, auto-analysis Launch (Human) Halt (Automatic) Assess (Automatic) (c) A human operator launches a script containing several fully automated test Batch Launch (Human) Launch (Automatic) Halt (Automatic) Assess (Automatic) Batch Assessment (Human Types of Analysis ====================================================== Single Pass Fail - Did an event happen when it was expected Perpetual Single Pass/Fail - Confirming an event that never should happen, never happened (e.g., multiple obstacle avoidance over time etc) Continuous Pass/Fail Counter - Checking a condition that will fail periodically and we are checking the rate of failure. Other data may be stored with each case. For example, did ownship "catch" an intruder? Also recording the intruder's average speed Countinuous Evaluation - Run until a certain criteria has been satisfied, e.g., endflag or max amount of time: then post a variable representing an evaluation Comparison to sister mission Types of Missions ==================================== o Single Vehicle o Multi-Vehicle, but focus on on rating a single ownship o Multi-Vehicle, rating is based on group performance Types of Tests ======================================================= A Test - launch - halt - assess A Batch Test - A set of tests - Batch results assessed Exhaustive Batch Test - Starting positions exhaustively tested - Autonomy parameters exhaustively tested - Sensor parameters Parameter Randomized Batch Test - Starting positions randomized - Autonomy parameters randomized - Sensor parameter Natural Randomized Batch Test - Hold starting positions, autonomy params and sensor params - Let natural system randomness vary When to Measure/Test Automated mission testing implies that the launch of the mission is Two types of Types of evaluations 1. Simple pass-fail logic evaluations o The result will be a Boolean o There is a single pass/fail check after a particular event Examples: A. Check that an endflag was posted at the end of a waypoint behavior This is typically to check that a certain kind of stated functionality, for a particular app or particular behavior, performs as state. 2. Complex pass-fail logic conditions o The result will be a Boolean o Multiple pass/fail checks are performed after multiple events Examples: o 3. o The result will be a Boolean o A vehicle runs for some period of time