more state vars

js/agent.js

@@ -22,7 +22,7 @@ function Agent(opt, world) {
 Agent.prototype.init = function (actor, critic) {
     var actions = this.walker.joints.length
     var temporal = 1
-    var states = this.walker.bodies.length * 2
+    var states = this.walker.bodies.length * 7
 
     var input = window.neurojs.Agent.getInputDimension(states, actions, temporal)
 
@@ -39,6 +39,9 @@ Agent.prototype.init = function (actor, critic) {
         temporalWindow: temporal, 
 
         discount: 0.97, 
+        rate: 0.004,
+        theta: 0.05, // progressive copy
+        alpha: 0.1, // advantage learning
 
         experience: 75e3, 
         // buffer: window.neurojs.Buffers.UniformReplayBuffer,
@@ -46,9 +49,7 @@ Agent.prototype.init = function (actor, critic) {
         learningPerTick: 40, 
         startLearningAt: 900,
 
-        theta: 0.05, // progressive copy
 
-        alpha: 0.1 // advantage learning
 
     })
 

js/game.js

@@ -27,7 +27,7 @@ gameInit = function() {
   var joints = 12
   var bodyParts = 14
   var sensors = 14
-  var state = sensors * 2
+  var state = sensors * 7
   var actions = joints
   var input = 2 * state + 1 * actions
 
@@ -81,11 +81,15 @@ logRewards = function () {
   }
 }
 
+
 resetSimulation = function () { 
+  // turn training off temporarlity to avoid NaN's
+  updateIfLearning(false)
   console.log('resetting walkers')
   for(var k = 0; k < config.population_size; k++) {
     globals.agents[k].walker = globals.walkers[k] = new Walker(globals.world)
   }
+  setTimeout(()=>updateIfLearning(true), 1000)
 }
 
 simulationStep = function() {
@@ -97,10 +101,6 @@ simulationStep = function() {
   } else {
     globals.step_counter++;
   }
-  // document.getElementById("generation_timer_bar").style.width = (100*globals.step_counter/config.round_length)+"%";
-  // if(globals.step_counter > config.round_length) {
-  //   nextGeneration();
-  // }
 }
 
 setSimulationFps = function(fps) {
@@ -122,17 +122,9 @@ setSimulationFps = function(fps) {
 }
 
 createPopulation = function(genomes) {
-  // setQuote();
-  // if(typeof globals.generation_count == 'undefined') {
-  //   globals.generation_count = 0;
-  // } else {
-  //   globals.generation_count++;
-  // }
-  // updateGeneration(globals.generation_count);
   var walkers = [];
   var agents = []
   for(var k = 0; k < config.population_size; k++) {
-    // walkers.push(new Walker(globals.world));
     var agent = new Agent({}, globals)
     agents.push(agent);
     walkers.push(agent.walker)
@@ -141,22 +133,11 @@ createPopulation = function(genomes) {
   return [agents, walkers];
 }
 
-randAction= function(){ 
-  var action = []
-  for (let i = 0; i < 12; i++) {
-    action.push(Math.randf(-1,1))
-  }
-  return action
-}
 
 populationSimulationStep = function() {
   for(var k = 0; k < config.population_size; k++) {
-      // var action = randAction()
-    // globals.walkers[k].simulationStep(action);
     globals.agents[k].step()
-    // console.log(globals.walkers[k].walker.getState())
-  }
-  
+  }  
 }
 
 
@@ -207,9 +188,7 @@ readBrain = function (buf) {
 
 
 updateIfLearning = function (value) {
-  for (var i = 0; i <  globals.world.agents.length; i++) {
-    globals.world.agents[i].brain.learning = value
+  for (var i = 0; i <  globals.agents.length; i++) {
+    globals.agents[i].brain.learning = value
   }
-
-  globals.world.plotRewardOnly = !value
 };