updated README and resolved merge conflict

README.md

@@ -8,7 +8,7 @@ This repository is a fork of [gym-pybullet-drones](https://github.com/utiasDSL/g
 
 - The drone can follow arbitrary trajectories. 
 - It is given two next target waypoints as observation. If the two target waypoints are close, it will reach the target slower. 
-- The learned policy is well-suited to smoothly follow arbitrary trajectories and corresponds to the obtained result after _slow-phase training_ in [Penicka et al. [1]](https://rpg.ifi.uzh.ch/docs/RAL_IROS22_Penicka.pdf).
+- The learned policy corresponds to the obtained result after _slow-phase training_ in [Penicka et al. [1]](https://rpg.ifi.uzh.ch/docs/RAL_IROS22_Penicka.pdf).
 
 ## Implemented New Features
 
@@ -19,9 +19,48 @@ This repository is a fork of [gym-pybullet-drones](https://github.com/utiasDSL/g
 
 ## Setup
 
+Tested on ArchLinux and Ubuntu. Note that Eigen must be installed on the system. On linux, install via your package manager. E.g. on Ubuntu:
+
+```s
+$ sudo apt-get install libeigen3-dev
+```
+
+It is strongly recommended to use a python virtual environment, such as _conda_ or _pyvenv_.
+
+1. Initialise repository. Repository must be pulled recursively
+
+```s
+$ git clone git@github.com:danielbinschmid/RL-pybullets-cf.git
+$ git submodule --init --recursive
+```
+
+2. Initialise virtual environment. Tested with python version 3.10.13. E.g.:
+
+```s
+$ pyenv install 3.10.13
+$ pyenv local 3.10.13
+$ python3 -m venv ./venv
+$ source ./venv/bin/activate
+$ pip3 install --upgrade pip
+```
+
+3. Install dependencies and build
+
+```s
+$ pip3 install -e . # if needed, `sudo apt install build-essential` to install `gcc` and build `pybullet`
+```
+
+## Usage
 
 ## Implementation Overview
 
+## Dev
+
+
+## Common Issues
+
+- *Existing and mismatching CMakeCache.txt in [trajectories/trajectories_generation](./trajectories/trajectories_generation).* Solution: Remove CMakeCache.txt in build folder of [trajectories/trajectories_generation](./trajectories/trajectories_generation).
+
 ## References 
 
 - [1]: Penicka, Robert, et al. [*Learning minimum-time flight in cluttered environments.*](https://rpg.ifi.uzh.ch/docs/RAL_IROS22_Penicka.pdf) IEEE Robotics and Automation Letters 7.3 (2022): 7209-7216.

agents/C_trajectory_following.py

@@ -28,20 +28,14 @@
 ####################################
 
 ###### TEST TRAIN FLAGS ############
-TRAIN = False
+TRAIN = True
 TEST = True
 ####################################
 
 ###### ENVIRONMENT PARAMS ##########
-<<<<<<< HEAD
-TIMESTEPS = 1e6
-N_ENVS = 5
-EPISODE_LEN_SEC = 6
-=======
 TIMESTEPS = 2.5e6
 N_ENVS = 20
 EPISODE_LEN_SEC = 20
->>>>>>> test_suite
 ####################################
 
 ###### HYPERPARAMS #################
@@ -91,14 +85,9 @@ def run(output_folder=OUTPUT_FOLDER,
     # CONFIG ##################################################
     t_traj, init_wp = init_targets()
 
-<<<<<<< HEAD
-    # output_folder = f"{output_folder}/k_p={k_p}_k_wp={k_wp}_k_s={k_s}_max_reward_distance={max_reward_distance}_waypoint_dist_tol={waypoint_dist_tol}"
-    # print(f"Output folder: {output_folder}")
-=======
     # random number in range 10-99
     output_folder = f"{output_folder}/wp_b={waypoint_buffer_size}_k_p={k_p}_k_wp={k_wp}_k_s={k_s}_max_reward_distance={max_reward_distance}_waypoint_dist_tol={waypoint_dist_tol}"
     print(f"Output folder: {output_folder}")
->>>>>>> test_suite
 
     config = Configuration(
         action_type=ACT,
@@ -129,11 +118,7 @@ def run(output_folder=OUTPUT_FOLDER,
                   env_factory=env_factory)
 
     if test:
-<<<<<<< HEAD
-        for _ in range(5):
-=======
         for _ in range(10):
->>>>>>> test_suite
             run_test(config=config,
                     env_factory=env_factory)