Skip to content

Env sweeps over battle and drone #650

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
vyeoms wants to merge 2 commits into
base: 5.0
Choose a base branch
from
+11 −4
Open

Env sweeps over battle and drone #650

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1c66bfb
Widen drone lower sweep range
vyeoms Aug 17, 2026
6bfd048
(battle) Fix perf+score logging
vyeoms Aug 17, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 6 additions & 0 deletions config/drone.ini
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -85,6 +85,12 @@ max = 2e8
mean = 8e7
scale = auto

[sweep.train.clip_coef]
distribution = uniform
min = 0.01
max = 0.9
scale = auto

# hover
[sweep.env.hover_alpha_dist]
distribution = log_normal
Expand Down
9 changes: 5 additions & 4 deletions ocean/battle/battle.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -767,9 +767,9 @@ void compute_observations(Battle* env) {
obs[obs_idx++] = agent->y;
obs[obs_idx++] = agent->z;
obs[obs_idx++] = agent->y - ground_height(env, agent->x, agent->z);
obs[obs_idx++] = abs(agent->x) - 0.95f*env->size_x;
obs[obs_idx++] = abs(agent->z) - 0.95f*env->size_z;
obs[obs_idx++] = abs(agent->y) - 0.95f*env->size_y;
obs[obs_idx++] = fabsf(agent->x) - 0.95f*env->size_x;
obs[obs_idx++] = fabsf(agent->z) - 0.95f*env->size_z;
obs[obs_idx++] = fabsf(agent->y) - 0.95f*env->size_y;
obs[obs_idx++] = agent->speed;
obs[obs_idx++] = agent->health;
obs[obs_idx++] = agent->max_turn;
Expand Down Expand Up @@ -921,7 +921,8 @@ void puf_step(Battle* env) {
if (i < env->num_agents/2) {
env->agents[i].rewards[0] = reward;
env->agents[i].terminals[0] = 1;
env->log.score = env->log.episode_return;
env->log.score += agent->episode_return;
env->log.perf += (agent->episode_return > 0.0f) ? 1 : 0;
env->log.episode_length += agent->episode_length;
env->log.episode_return += agent->episode_return;
env->log.collision_rate += collision;
Expand Down
Loading