Scatter plot to compare mean particles vs lmfit parameters

scripts/helpers.jl

@@ -2,21 +2,72 @@
 helpers.jl
 This file provides helper functions to experiments.jl, the experiment script
 Most functions have the script within the doc string as example
-
-# List of functions
-- metrics_from_jld
-- metrics_from_jld_idmbased
-- multiscenarios_idm
-- multiscenarios_pf
-- collision rmse functions
-- scenario generation
 """
 
 using Distributions # provides `mean`: to compute mean of particle dist over cars
 using JLD # to save models
 using AutomotiveInteraction
 using DelimitedFiles # to write to txt file that will be read in by python
 
+"""
+- Make scatter plot of the lmidm parameters and final particle for a scenario
+with each scatter point corresponding to a different car
+
+# Examples
+```julia
+cd("scripts")
+f= FilteringEnvironment()
+scatterplot_lmfit_pf(f,scene_real,scenario_name="upper_1")
+```
+"""
+function scatterplot_lmfit_pf(f::FilteringEnvironment;scenario_name="upper_1")
+    filename="media/$(scenario_name).jld"
+    particle,id_list,ts,te = JLD.load(filename,"p","veh_id_list","ts","te")
+    scene = deepcopy(f.traj[ts])
+    if !isempty(id_list) keep_vehicle_subset!(scene,id_list) end
+
+    v_des_vec = []
+    T_vec = []
+    v_des_vec_pf = []
+    T_vec_pf=[]
+    # Loop over list of vehicles and assign model params based on lmfit values
+    for veh in scene
+        veh_id = veh.id
+
+        # Read the lmfit parameters
+        lmfit_param_filename = "lmfit/$(scenario_name)/$(veh_id)_lmfit_params.txt"
+        if isfile(lmfit_param_filename)
+            lmparams = readdlm(lmfit_param_filename)
+            v_des = lmparams[1]
+            T = lmparams[2]
+            s_min = lmparams[3]
+            push!(v_des_vec,v_des)
+            push!(T_vec,T)
+        else
+            # Default to Jeremy's parameters
+            print("Could not find lmfit param file for $(veh_id). Use Jeremy params\n")
+            v_des=17.837
+            T=0.918
+            s_min=5.249
+            push!(v_des_vec,v_des)
+            push!(T_vec,T)
+        end
+
+        # Insert the pf parameters
+        push!(v_des_vec_pf,particle[veh_id][1])
+        push!(T_vec_pf,particle[veh_id][3])
+    end
+
+    scatter_lmidm = PGFPlots.Plots.Scatter(Float64.(v_des_vec),Float64.(T_vec),
+    legendentry="lmidm")
+    scatter_pf = PGFPlots.Plots.Scatter(Float64.(v_des_vec_pf),Float64.(T_vec_pf),
+    legendentry="pf")
+    a = PGFPlots.Axis([scatter_lmidm,scatter_pf],xlabel="v",ylabel="T",
+    title="lmidm vs pf parameters")
+    PGFPlots.save("media/lmfit_pf_scatter_$(scenario_name).svg",a)
+    return nothing
+end
+
 #****************scenario_generation************************
 """
 function train_one_test_another(;train_filename="media/lower_3.jld",test_filename="media/upper_2.jld",video_filename="media/train_low_test_up.mp4")

src/Driving/driving_simulation.jl

@@ -157,7 +157,8 @@ function make_cidm_models(f::FilteringEnvironment,scene)
     print("c-IDM models being assigned to vehicles\n")
     models = Dict{Int64,DriverModel}()
     for veh in scene
-        models[veh.id] = CooperativeIDM(env=f.mergeenv,c=1.0)
+        models[veh.id] = CooperativeIDM(env=f.mergeenv,c=1.0,
+                        idm=IntelligentDriverModel(v_des=5.))
     end
     return models
 end

src/Filtering/metrics.jl

@@ -209,16 +209,14 @@ end
 
 # Example
 ```julia
-rmse_pos_mat_idm,rmse_vel_mat_idm,coll_mat_idm = 
-        multiscenarios_idm(mergetype="upper",modelmaker=make_IDM_models)
-rmse_pos_mat_cidm, rmse_vel_mat_cidm, coll_mat_cidm = 
-        multiscenarios_idm(mergetype="upper",modelmaker=make_cidm_models)
-rmse_pos_mat_lmidm, rmse_vel_mat_lmidm, coll_mat_lmidm = multiscenarios_lmidm(
-    mergetype="upper")
-rmse_pos_mat_pf, rmse_vel_mat_pf, coll_mat_pf = multiscenarios_pf(mergetype="upper")
-
-coll_mat_list = [coll_mat_idm,coll_mat_cidm,coll_mat_lmidm,coll_mat_pf]
-coll_barchart(coll_mat_list,filename = "media/coll_barchart_new.svg")
+cd("scripts")
+rmse_pos_mat_idm,rmse_vel_mat_idm,coll_mat_idm = multiscenarios_idm(mergetype="upper",modelmaker=make_IDM_models);
+rmse_pos_mat_cidm, rmse_vel_mat_cidm, coll_mat_cidm = multiscenarios_idm(mergetype="upper",modelmaker=make_cidm_models);
+rmse_pos_mat_lmidm, rmse_vel_mat_lmidm, coll_mat_lmidm = multiscenarios_lmidm(mergetype="upper");
+rmse_pos_mat_pf, rmse_vel_mat_pf, coll_mat_pf = multiscenarios_pf(mergetype="upper");
+
+coll_mat_list = [coll_mat_idm,coll_mat_cidm,coll_mat_lmidm,coll_mat_pf];
+coll_barchart(coll_mat_list,filename = "media/coll_barchart_new_cidm.svg");
 ```
 """
 function coll_barchart(coll_mat_list;filename="media/test_bar.pdf")
@@ -246,13 +244,14 @@ function rmse_plots_modelscompare(rmse_list;filename="media/test_rmse.pdf")
 
 # Example
 ```julia
-rmse_pos_mat_idm,rmse_vel_mat_idm,coll_mat_idm = multiscenarios_idm(mergetype="upper",modelmaker=make_IDM_models)
-rmse_pos_mat_cidm, rmse_vel_mat_cidm, coll_mat_cidm = multiscenarios_idm(mergetype="upper",modelmaker=make_cidm_models)
-rmse_pos_mat_lmidm, rmse_vel_mat_lmidm, coll_mat_lmidm = multiscenarios_lmidm(mergetype="upper")
-rmse_pos_mat_pf, rmse_vel_mat_pf, coll_mat_pf = multiscenarios_pf(mergetype="upper")
-
-rmse_list = [rmse_pos_mat_idm,rmse_pos_mat_cidm,rmse_pos_mat_lmidm,rmse_pos_mat_pf]
-rmse_plots_modelscompare(rmse_list,filename = "media/rmse_pos_new.svg")
+cd("scripts")
+rmse_pos_mat_idm,rmse_vel_mat_idm,coll_mat_idm = multiscenarios_idm(mergetype="upper",modelmaker=make_IDM_models);
+rmse_pos_mat_cidm, rmse_vel_mat_cidm, coll_mat_cidm = multiscenarios_idm(mergetype="upper",modelmaker=make_cidm_models);
+rmse_pos_mat_lmidm, rmse_vel_mat_lmidm, coll_mat_lmidm = multiscenarios_lmidm(mergetype="upper");
+rmse_pos_mat_pf, rmse_vel_mat_pf, coll_mat_pf = multiscenarios_pf(mergetype="upper");
+
+rmse_list = [rmse_pos_mat_idm,rmse_pos_mat_cidm,rmse_pos_mat_lmidm,rmse_pos_mat_pf];
+rmse_plots_modelscompare(rmse_list,filename = "media/rmse_pos_new.svg");
 ```
 """
 function rmse_plots_modelscompare(rmse_mat_list;filename="media/test_rmse.pdf")