0.25
0.5
0.75
1.25
1.5
1.75
2
Reachability and Learning for Hybrid Systems
Published on Jul 28, 20152622 Views
Hybrid systems are a modeling tool allowing for the composition of continuous and discrete state dynamics. They can be represented as continuous systems with modes of operation modeled by discrete dyn
Related categories
Chapter list
Reachability and Learning for Hybrid Systems00:00
Air traffic in Oakland Center00:28
Controller must keep aircraft separated - 101:39
Controller must keep aircraft separated - 202:40
Growing numbers of UAV applications03:15
Hybrid System Model04:18
Outline - 105:42
Backwards Reachable Set06:02
A discrete game - 107:24
A discrete game - 208:28
A discrete game - 309:39
A discrete game - 409:55
Reachable Set Propagation - 110:11
Reachable Set Propagation - 211:44
Reachable Set Propagation - 311:53
Level set interpretation - 111:58
Level set interpretation - 212:25
Level set interpretation - 312:50
Level set interpretation - 412:53
Level set interpretation - 513:00
Partitions the state space13:18
Example 1: Collision Avoidance13:35
Backwards Reachable Set: Capture14:37
Mode sequencing and reach-avoid15:00
Dealing with the curse of dimensionality16:18
Outline - 216:50
Example 2: Platooning UAVs - 116:51
Example 2: Platooning UAVs - 217:52
Merging onto highway and joining platoon - 118:27
Merging onto highway and joining platoon - 219:07
Intruder vehicle19:24
Example 3: Forced Landing System19:43
Outline - 320:54
Learn models from data …22:29
Example 4: Safe - Policy Gradient Reinforcement Learning - 123:07
Example 4: Safe - Policy Gradient Reinforcement Learning - 224:19
Gaussian Processes (GP)25:39
Online Disturbance Model Validation - 126:14
Online Disturbance Model Validation - 226:55
Online Disturbance Model Validation - 327:11
Example 5: Safe Learning - 127:20
Example 5: Safe Learning - 227:41
Local Updates28:06
ocal Updates using Temporal Difference Learning - 129:00
ocal Updates using Temporal Difference Learning - w230:01
Example 6: Learning to Fly (in a confined space with unknown payload)30:18
Conclusions and current work - 131:27
Conclusions and current work - 232:24
Thanks33:06