350 likes | 510 Views
Micro-Aerial Vehicle Swarms. Karthik Dantu Bryan Kate Jason Waterman Peter Bailis Matt Welsh. Programming With Karma. Presentors: Yuxuan Dai, Long Ma. MAV Swarms Background. Advantages. Can be applied to enclosed, close-quarter areas. Extremely small.
E N D
Micro-Aerial Vehicle Swarms Karthik Dantu Bryan Kate Jason Waterman Peter Bailis Matt Welsh ProgrammingWith Karma Presentors: Yuxuan Dai, Long Ma
MAV Swarms Background Advantages Can be applied to enclosed, close-quarter areas Extremely small deployed to achieve a task massively Can perform tasks that challenging for larger platforms Large number Provide robustness to failure in the field
MAV Swarms Background Restrictions Limited resources Complex coordination swarm size
Karma system Simplify MAVs’ behavior uniformly Combine MAV behaviors with a swarm-level goal Eliminate low-level MAV coordination from users Karma system
Karma system introduction (X, Y) programming model hive-drone model assumption restriction drones cannot communicate with each other drones operate without precise knowledge of location
Content Introduction Hive-Drone Model Karma Implementation Evaluation Our Thoughts
Hive-Drone Model Coordination complexity restriction restriction Programmer cannot explicitly coordinate MAV’s behavior drones cannot communicate with each other
Sortie Disease! Behavior Application
Disease! Spatial Decomposition 1. Make it easier to reason about MAV allocation. 2. It is unlikely that the MAVs can access to high-resolution location services in the field 3. Using a Cartesian coordinate system is not necessary
T,R,N T,R,N Data Model The hive maintains a key-value repository called the Datastore. Updates to this data structure are asynchronous, occurring when drones return from a sortie The Datastore can be queried both temporally and spatially
Programming Model Every behavior produces some type of information under normal execution Based on the information in the Datastore ! 1. Activation Predicate 2. Progress Function
Scheduling Problem Drone 4 run behaviors Ain R3 (t4) Drone 1 run behaviors Ain R1 (t1) 1. Use the shortest time to Complete application Advocate scheduling all available drones greedily The hive-drone, programming model transform the problem of executing an application on a MAV swarm into a problem of scheduling behaviors on drones. Drone 3 run behaviors Ain R7 (t3) Drone 2 run behaviors Bin R6 (t2) System executes behaviors that are concurrently activated in sequence
If behavior A and B are activated concurrently (greedily schedule all available drones) Interleaves allocations for A and B Schedule all drones for A then for B No distinction Drone N run behaviors Ain Rx (tn) Drone 1 run behaviors Ain R1 (t1) Drone n run behaviors Bin Rx (tn) Drone 1 run behaviors Bin R1 (t2)
Scheduling Problem 1. Use the shortest time to Complete application 2. Achieve fairness between behaviors output Minimize the difference in progress between any two activated behaviors
Content Introduction Hive-Drone Model Karma Implementation Evaluation Our Thoughts
Karma Implementation Karma Controller Karma Implementation Scheduler Dispatcher Datastore
Process: a. Estimate the total workload b. Allocate the available drones fairly basing the remaining amount of work.
Process: a. Help drove transmit the data to hive Datastore b. Charge the droves c. Notify the Controller of the resource availability
Karma Implementation Two kinds of Dispatch policies: a. Continuous dispatch policy Provide a constant presence of drones in the field Minimize the information latency b. Greedy dispatch policy Dispatch the drones opportunistically
Content Introduction Hive-Drone Model Karma Implementation Evaluation Our Thoughts
Evaluation Evaluate the effectiveness from three aspects: Execution time Energy cost Information latency
Evaluation Karma: a. greedy dispatching b. continuous dispatching Oracle: with foreknowledge of all activities, hence it can give a lower bound of requirements.
Evaluation Resilience to Failure fail-->cannot update Datastore-->Dispatcher detects that-->Scheduler re-arranges in next allocation cycle. Can be mitigated by reserving drones or increasing the swarm size.
Evaluation Adaptability introduce a constant wind over the bottom third of the field. In that field, the round trip time reduces 32% For same amount of work, 12% more drones. Equally, 7% higher energy cost
Evaluation • Hive-drone paradigm can be used to continuously measure time varying phenomena • Example: a chemical plume tracking application Interesting feature: static VS active
Active Sliding window T:5m Static
Our Thoughts • The current system is limited by the flight time of drones. If the drones had a longer flight time, they can operate different tasks per sortie which makes the system more efficient. • The assumption in the design is that the drones cannot communicate with each other. If the communication could be achieved, the allocation policy might be changed and information latency might also be reduced. • Inspired by idea of communication between two drones, communication between two hives may improve the coverage of application.