Got threading and SW limits working really well. Still might be some shakiness on deinitialization (need to coordinate GPIO.cleanup with threads), but program does exit and return home normally. Swingup test now runs very smooth and SW limits were increased with the drastically improved response time. Pushing to merge back to master.

1 files changed, 9 insertions, 6 deletions

diff --git a/System/system_swingup_test_2.py b/System/system_swingup_test_2.py
index e240b01..5c0ac5a 100644
--- a/System/system_swingup_test_2.py
+++ b/System/system_swingup_test_2.py
@@ -53,13 +53,13 @@ class SwingUpEnv():
     }

 

     def __init__(self):

-        self.sys = System(angular_units='Radians', positive_limit=10., negative_limit=-10., sw_limit_routine=self.x_threshold_routine)

+        self.x_threshold = 14.

+        self.sys = System(angular_units='Radians', positive_limit=self.x_threshold, negative_limit=-self.x_threshold, sw_limit_routine=self.x_threshold_routine)

 		

-        self.force_mag = 10.

+        self.force_mag = 11.

         self.last_time = time.time()  # time for seconds between updates

 

         # Angle at which to fail the episode

-        self.x_threshold = 10.

         self.x_dot_threshold = 10.

         self.theta_dot_threshold = 3*np.pi

 

@@ -134,6 +134,7 @@ class SwingUpEnv():
         return np.array(self.state), reward, done, {'max_up_time' : self.max_up_time}

 

     def x_threshold_routine(self):

+        print("SW Limit reached!")

         self.done = True

         self.sys.adjust(0)

 

@@ -311,14 +312,16 @@ class sarsaAgent:
 env = SwingUpEnv()

 

 # For simplicity, we only consider forces of -1 and 1

-numActions = 5

+numActions = 2

 Actions = np.linspace(-1, 1, numActions)

 

 # This is our learning agent

 gamma = .95

-agent = sarsaAgent(5, numActions, 20, 1, epsilon = 5e-2, gamma = gamma, alpha = 1e-5)

+#agent = sarsaAgent(5, numActions, 20, 1, epsilon = 5e-2, gamma = gamma, alpha = 1e-5)

+agent = deepQagent(5,numActions,20,2,epsilon=5e-2,gamma=gamma,batch_size=20,

+                   c= 100,alpha=1e-4)

 

-maxSteps = 5e4

+maxSteps = 2e5

 

 # This is a helper to deal with the fact that x[2] is actually an angle

 x_to_y = lambda x : np.array([x[0], x[1], np.cos(x[2]), np.sin(x[2]), x[3]])