@@ -0,0 +1,102 @@
+# Copyright (c) 2026 Uwe Fechner
+# SPDX-License-Identifier: MPL-2.0
+
+"""
+Extra plots for the LearningControl paper that are not part of the pattern
+figures of `plot_patterns_paper.jl`.
+
+- `plot_attractor_distance`: the arc length [m] from the closest path point Q to
+  the attractor point over time, for the scenarios in
+  [`ATTRACTOR_SCENARIOS`](@ref) (Cabauw and Maasvlakte at low, medium and high
+  wind speed) in one plot. Saved as `attractor_distance.pdf` into
+  `../LearningControl/figures`.
+
+    include("plots_extra.jl")
+"""
+
+using Pkg
+if Base.active_project() != joinpath(@__DIR__, "Project.toml")
+    Pkg.activate(joinpath(@__DIR__))
+end
+
+using GLMakie
+using MakieControlPlots
+using V3Kite
+using SimpleKiteControllers
+
+include(joinpath(@__DIR__, "plot_pattern_utils.jl"))
+
+const PAPER_FIGURES_DIR = normpath(joinpath(@__DIR__, "..", "..", "LearningControl", "figures"))
+
+"""
+    ATTRACTOR_SCENARIOS
+
+`(site, scenario)` pairs drawn by [`plot_attractor_distance`](@ref): both sites
+at 4, 7 and 10 m/s.
+"""
+const ATTRACTOR_SCENARIOS = [
+    ("cabauw", "v04"), ("cabauw", "v07"), ("cabauw", "v10"),
+    ("maasvlakte", "v04"), ("maasvlakte", "v07"), ("maasvlakte", "v10"),
+]
+
+"""
+    attractor_arc(site, scenario) -> (time, arc, flown_wind)
+
+The attractor arc `Δ·L` [m] of the run in `output/scenarios/<site>/<scenario>`
+over time: [`attractor_distance`](@ref) evaluated with the scenario's own
+`fc_settings_reelout.yaml` at the logged apparent wind speed and tether length,
+the same value the guidance re-reads every step. `NaN` outside the phases
+transition, fig8 and final (3–5); in the entry phases before them the state
+machine replaces the guidance. `flown_wind` is the ground wind speed [m/s].
+"""
+function attractor_arc(site, scenario)
+    scenario_dir = normpath(joinpath(@__DIR__, "..", "output", "scenarios", site, scenario))
+    isdir(scenario_dir) || error("$scenario_dir does not exist.")
+    fcs = FC_Settings(fc_settings(scenario_system_file(scenario_dir)); path = scenario_dir)
+
+    log_name = replace(only(filter(f -> endswith(f, ".arrow"), readdir(scenario_dir))), ".arrow" => "")
+    sl = load_log(log_name; path = scenario_dir).syslog
+    summary_sim = V3Kite.YAML.load_file(joinpath(scenario_dir, log_name * ".yaml"))["simulation"]
+
+    rng = 2:length(sl.time)
+    time = Float64.(sl.time[rng])
+    l_tether = getindex.(sl.l_tether[rng], 1)
+    v_app = Float64.(sl.v_app[rng])
+    guided = sl.sys_state[rng] .>= 3
+    arc = [guided[i] ? deg2rad(attractor_distance(fcs, v_app[i], l_tether[i])) * l_tether[i] : NaN
+           for i in eachindex(time)]
+    return time, arc, summary_sim["wind_speed"]
+end
+
+"""
+    plot_attractor_distance(scenarios = ATTRACTOR_SCENARIOS; t_max = 180.0, disp = true, save = true)
+
+Plot the attractor arc ([`attractor_arc`](@ref)) of each `(site, scenario)` pair
+in `scenarios` over time, one line per run. The runs share one time axis, that of
+the longest run; the others are sampled onto it, `NaN` beyond their own end.
+The time axis runs from 0 to `t_max` [s].
+"""
+function plot_attractor_distance(scenarios = ATTRACTOR_SCENARIOS; t_max = 180.0, disp = true, save = true)
+    runs = [attractor_arc(site, scenario) for (site, scenario) in scenarios]
+    time = argmax(r -> last(r[1]), runs)[1]
+    arcs = map(runs) do (t_run, arc, _)
+        map(time) do t
+            k = searchsortedlast(t_run, t)
+            k == 0 || t > last(t_run) ? NaN : arc[k]
+        end
+    end
+    labels = ["$(uppercasefirst(site)) $(round(r[3]; digits = 1)) m/s"
+              for ((site, _), r) in zip(scenarios, runs)]
+
+    p = MakieControlPlots.plot(time, arcs;
+        xlabel = "time [s]", ylabel = "attractor distance [m]",
+        labels = labels, xlims = (0.0, t_max), fig = "attractor distance", disp = disp)
+    if save && disp
+        pdf_file = joinpath(PAPER_FIGURES_DIR, "attractor_distance.pdf")
+        savefig(pdf_file)
+        @info "Saved attractor distance plot" pdf_file
+    end
+    return p
+end
+
+plot_attractor_distance()