cleanup

Author: Yuxiao Chen

README.md

@@ -26,7 +26,7 @@ pip install -e .
 Install `Pplan`
 ```
 cd ..
-git clone ssh://[email protected]:NVlabs/policy-planner.git Pplan
+git clone ssh://[email protected]:NVlabs/spline-planner.git Pplan
 cd Pplan
 pip install -e .
 ```

setup.py

@@ -26,7 +26,6 @@
         "imageio-ffmpeg",
         "python-louvain",
         "networkx",
-        "casadi",
         "torchtext",  # weird pytorch-lightning dependency bug
         "pytorch-lightning",
         "nuscenes-devkit",

tbsim/envs/env_trajdata.py

@@ -743,7 +743,6 @@ def _step(self, step_actions: RolloutAction, num_steps_to_take):
         # obs = {k:v for k,v in obs.items() if not isinstance(v,list)}
 
         # record metrics
-        #TODO: fix the bugs in metrics when using diffstack
         self._add_per_step_metrics(obs)
         if step_actions.has_ego:
             combined_step_actions = self.combine_action(step_actions)

tbsim/models/rasterized_models.py

@@ -1,4 +1,3 @@
-import pdb
 from turtle import forward
 from typing import Dict, List
 from collections import OrderedDict
@@ -1854,22 +1853,6 @@ def obtain_pred(self,static_img,seq_img,curr_state,raster_pos):
 
         prob = prob.nan_to_num(nan=0.0, posinf=1.0, neginf=0.0)
         prob = prob/prob.sum(-1,keepdim=True)
-
-        # import matplotlib.pyplot as plt
-        # img1 = static_img[:,:3].cpu().numpy()
-        # img2 = seq_img.abs().max(1,keepdim=True)[0].cpu().numpy()
-        # img3 = torch.cat((static_img[:,:2],seq_img.abs().max(1,keepdim=True)[0]),1).cpu().numpy()
-        
-        # for i in range(seq_img.shape[1]):
-        #     img = torch.cat((static_img[:,:2],seq_img[:,i:i+1]),1).cpu().numpy()
-        #     fig,axes = plt.subplots(1,2)
-        #     axes[0].imshow(img[0].transpose(1,2,0))
-        #     axes[1].imshow(img[1].transpose(1,2,0))
-        #     plt.show()
-        # import pdb
-        # pdb.set_trace()
-        # pos_xy_rel = raster_pos.unsqueeze(1)/(seq_img.shape[-1]-1)*2-1
-        # xx = torch.nn.functional.grid_sample(seq_img,pos_xy_rel,align_corners=True,)
 
         res = self.postprocessor(pred_traj,raster_pos,curr_state)
 
@@ -2176,10 +2159,7 @@ def forward(self, batch: dict,predict=False,**kwargs):
         batched_pred["target_trajectory"] = scene_fut_traj[...,:self.stage*self.num_frames_per_stage,:]
 
         batched_pred["agent_avail"] = agent_avail.any(-1)
-        
-        # import pdb
-        # pdb.set_trace()
-        # print(torch.cuda.memory_allocated()/1024/1024/1024) 
+
         return batched_pred
 
     def _batching_from_stages(self,preds):

tbsim/models/unet.py

@@ -261,8 +261,6 @@ def _apply_encoder(self, x, s):
         e_in = self._e0(x)
         # Run all encoder blocks
         skip_out_dict = {}
-        # import pdb
-        # pdb.set_trace()
         for i in range(self._num_resnet_blocks):
             block = getattr(self, '_e{}'.format(i + 1))
             e_out = block(e_in)
@@ -501,8 +499,6 @@ def main():
                                                     )
     pred_logits, pred_traj = out
     traj = postprocessor(pred_traj,query_pts,curr_state)
-    import pdb
-    pdb.set_trace()