Use ndv2

The SNPE model check function now always returns False, bypassing its original logic. Additionally, the sunnypilot SConscript has been commented out, likely to exclude it from the build process. These changes could be temporary or part of a refactoring strategy.

SConstruct

@@ -396,7 +396,7 @@ SConscript(['third_party/SConscript'])
 
 SConscript(['selfdrive/SConscript'])
 
-SConscript(['sunnypilot/SConscript'])
+# SConscript(['sunnypilot/SConscript'])
 
 if Dir('#tools/cabana/').exists() and GetOption('extras'):
   SConscript(['tools/replay/SConscript'])

selfdrive/modeld/modeld.py

@@ -1,21 +1,13 @@
 #!/usr/bin/env python3
-import os
 from openpilot.system.hardware import TICI
 
+from openpilot.selfdrive.modeld.runners.model_runner import ONNXRunner, TinygradRunner
+
 #
-if TICI:
-  from tinygrad.tensor import Tensor
-  from tinygrad.dtype import dtypes
-  from openpilot.selfdrive.modeld.runners.tinygrad_helpers import qcom_tensor_from_opencl_address
-  os.environ['QCOM'] = '1'
-else:
-  from openpilot.selfdrive.modeld.runners.ort_helpers import make_onnx_cpu_runner
 import time
-import pickle
 import numpy as np
 import cereal.messaging as messaging
 from cereal import car, log
-from pathlib import Path
 from setproctitle import setproctitle
 from cereal.messaging import PubMaster, SubMaster
 from msgq.visionipc import VisionIpcClient, VisionStreamType, VisionBuf
@@ -33,13 +25,8 @@ from openpilot.selfdrive.modeld.fill_model_msg import fill_model_msg, fill_pose_
 from openpilot.selfdrive.modeld.constants import ModelConstants
 from openpilot.selfdrive.modeld.models.commonmodel_pyx import DrivingModelFrame, CLContext
 
-
 PROCESS_NAME = "selfdrive.modeld.modeld"
-SEND_RAW_PRED = os.getenv('SEND_RAW_PRED')
 
-MODEL_PATH = Path(__file__).parent / 'models/supercombo.onnx'
-MODEL_PKL_PATH = Path(__file__).parent / 'models/supercombo_tinygrad.pkl'
-METADATA_PATH = Path(__file__).parent / 'models/supercombo_metadata.pkl'
 
 class FrameMeta:
   frame_id: int = 0
@@ -61,35 +48,25 @@ class ModelState:
     self.prev_desire = np.zeros(ModelConstants.DESIRE_LEN, dtype=np.float32)
     self.full_features_20Hz = np.zeros((ModelConstants.FULL_HISTORY_BUFFER_LEN, ModelConstants.FEATURE_LEN), dtype=np.float32)
     self.desire_20Hz =  np.zeros((ModelConstants.FULL_HISTORY_BUFFER_LEN + 1, ModelConstants.DESIRE_LEN), dtype=np.float32)
+    # Initialize model runner
+    self.model_runner = TinygradRunner(self.frames) if TICI else ONNXRunner(self.frames)
 
     # img buffers are managed in openCL transform code
-    self.numpy_inputs = {
-      'desire': np.zeros((1, (ModelConstants.HISTORY_BUFFER_LEN+1), ModelConstants.DESIRE_LEN), dtype=np.float32),
-      'traffic_convention': np.zeros((1, ModelConstants.TRAFFIC_CONVENTION_LEN), dtype=np.float32),
-      'features_buffer': np.zeros((1, ModelConstants.HISTORY_BUFFER_LEN,  ModelConstants.FEATURE_LEN), dtype=np.float32),
-    }
+    self.numpy_inputs = {}
 
-    with open(METADATA_PATH, 'rb') as f:
-      model_metadata = pickle.load(f)
-    self.input_shapes =  model_metadata['input_shapes']
+    for key, shape in self.model_runner.input_shapes.items():
+      if key not in self.frames: # Managed by opencl
+        self.numpy_inputs[key] = np.zeros(shape, dtype=np.float32)
 
-    self.output_slices = model_metadata['output_slices']
-    net_output_size = model_metadata['output_shapes']['outputs'][1]
-    self.output = np.zeros(net_output_size, dtype=np.float32)
     self.parser = Parser()
 
-    if TICI:
-      self.tensor_inputs = {k: Tensor(v, device='NPY').realize() for k,v in self.numpy_inputs.items()}
-      with open(MODEL_PKL_PATH, "rb") as f:
-        self.model_run = pickle.load(f)
-    else:
-      self.onnx_cpu_runner = make_onnx_cpu_runner(MODEL_PATH)
+    net_output_size = self.model_runner.model_metadata['output_shapes']['outputs'][1]
+    self.output = np.zeros(net_output_size, dtype=np.float32)
 
-  def slice_outputs(self, model_outputs: np.ndarray) -> dict[str, np.ndarray]:
-    parsed_model_outputs = {k: model_outputs[np.newaxis, v] for k,v in self.output_slices.items()}
-    if SEND_RAW_PRED:
-      parsed_model_outputs['raw_pred'] = model_outputs.copy()
-    return parsed_model_outputs
+    num_elements = self.numpy_inputs['features_buffer'].shape[1]
+    step_size = int(-100 / num_elements)
+    self.full_features_20Hz_idxs = np.arange(step_size, step_size * (num_elements + 1), step_size)[::-1]
+    self.desire_reshape_dims = (self.numpy_inputs['desire'].shape[0], self.numpy_inputs['desire'].shape[1], -1, self.numpy_inputs['desire'].shape[2])
 
   def run(self, buf: VisionBuf, wbuf: VisionBuf, transform: np.ndarray, transform_wide: np.ndarray,
                 inputs: dict[str, np.ndarray], prepare_only: bool) -> dict[str, np.ndarray] | None:
@@ -100,36 +77,42 @@ class ModelState:
 
     self.desire_20Hz[:-1] = self.desire_20Hz[1:]
     self.desire_20Hz[-1] = new_desire
-    self.numpy_inputs['desire'][:] = self.desire_20Hz.reshape((1,25,4,-1)).max(axis=2)
+    self.numpy_inputs['desire'][:] = self.desire_20Hz.reshape(self.desire_reshape_dims).max(axis=2)
+
+    for key in self.numpy_inputs:
+      if key in inputs and key not in ['desire']:
+        self.numpy_inputs[key][:] = inputs[key]
 
-    self.numpy_inputs['traffic_convention'][:] = inputs['traffic_convention']
     imgs_cl = {'input_imgs': self.frames['input_imgs'].prepare(buf, transform.flatten()),
                'big_input_imgs': self.frames['big_input_imgs'].prepare(wbuf, transform_wide.flatten())}
 
-    if TICI:
-      # The imgs tensors are backed by opencl memory, only need init once
-      for key in imgs_cl:
-        if key not in self.tensor_inputs:
-          self.tensor_inputs[key] = qcom_tensor_from_opencl_address(imgs_cl[key].mem_address, self.input_shapes[key], dtype=dtypes.uint8)
-    else:
-      for key in imgs_cl:
-        self.numpy_inputs[key] = self.frames[key].buffer_from_cl(imgs_cl[key]).reshape(self.input_shapes[key])
+    # Prepare inputs using the model runner
+    self.model_runner.prepare_inputs(imgs_cl, self.numpy_inputs)
 
     if prepare_only:
       return None
 
-    if TICI:
-      self.output = self.model_run(**self.tensor_inputs).numpy().flatten()
-    else:
-      self.output = self.onnx_cpu_runner.run(None, self.numpy_inputs)[0].flatten()
-
-    outputs = self.parser.parse_outputs(self.slice_outputs(self.output))
+    # Run model inference
+    self.output = self.model_runner.run_model()
+    outputs = self.parser.parse_outputs(self.model_runner.slice_outputs(self.output), self.numpy_inputs.keys())
 
     self.full_features_20Hz[:-1] = self.full_features_20Hz[1:]
     self.full_features_20Hz[-1] = outputs['hidden_state'][0, :]
 
-    idxs = np.arange(-4,-100,-4)[::-1]
-    self.numpy_inputs['features_buffer'][:] = self.full_features_20Hz[idxs]
+    self.numpy_inputs['features_buffer'][:] = self.full_features_20Hz[self.full_features_20Hz_idxs]
+
+    if "desired_curvature" in outputs:
+      input_name_prev = None
+
+      if "prev_desired_curvs" in self.numpy_inputs.keys():
+        input_name_prev = 'prev_desired_curvs'
+      elif "prev_desired_curv" in self.numpy_inputs.keys():
+        input_name_prev = 'prev_desired_curv'
+
+      if input_name_prev is not None:
+        len = outputs['desired_curvature'][0].size
+        self.numpy_inputs[input_name_prev][0, :-len, 0] = self.numpy_inputs[input_name_prev][0, len:, 0]
+        self.numpy_inputs[input_name_prev][0, -len:, 0] = outputs['desired_curvature'][0]
     return outputs
 
 
@@ -190,7 +173,6 @@ def main(demo=False):
   meta_main = FrameMeta()
   meta_extra = FrameMeta()
 
-
   if demo:
     CP = get_demo_car_params()
   else:
@@ -272,6 +254,9 @@ def main(demo=False):
       'traffic_convention': traffic_convention,
       }
 
+    if "lateral_control_params" in model.numpy_inputs.keys():
+      inputs['lateral_control_params'] = np.array([sm["carState"].vEgo, steer_delay], dtype=np.float32)
+
     mt1 = time.perf_counter()
     model_output = model.run(buf_main, buf_extra, model_transform_main, model_transform_extra, inputs, prepare_only)
     mt2 = time.perf_counter()

selfdrive/modeld/parse_model_outputs.py

@@ -84,7 +84,8 @@ class Parser:
     outs[name] = pred_mu_final.reshape(final_shape)
     outs[name + '_stds'] = pred_std_final.reshape(final_shape)
 
-  def parse_outputs(self, outs: dict[str, np.ndarray]) -> dict[str, np.ndarray]:
+  def parse_outputs(self, outs: dict[str, np.ndarray], input_keys: [str]) -> dict[str, np.ndarray]:
+    """ Parse the model outputs into a dictionary of numpy arrays. The input_keys are used to determine how the output should be parsed. """
     self.parse_mdn('plan', outs, in_N=ModelConstants.PLAN_MHP_N, out_N=ModelConstants.PLAN_MHP_SELECTION,
                    out_shape=(ModelConstants.IDX_N,ModelConstants.PLAN_WIDTH))
     self.parse_mdn('lane_lines', outs, in_N=0, out_N=0, out_shape=(ModelConstants.NUM_LANE_LINES,ModelConstants.IDX_N,ModelConstants.LANE_LINES_WIDTH))
@@ -96,6 +97,8 @@ class Parser:
                    out_shape=(ModelConstants.LEAD_TRAJ_LEN,ModelConstants.LEAD_WIDTH))
     if 'lat_planner_solution' in outs:
       self.parse_mdn('lat_planner_solution', outs, in_N=0, out_N=0, out_shape=(ModelConstants.IDX_N,ModelConstants.LAT_PLANNER_SOLUTION_WIDTH))
+    if 'desired_curvature' in outs and "prev_desired_curv" in input_keys:
+      self.parse_mdn('desired_curvature', outs, in_N=0, out_N=0, out_shape=(ModelConstants.DESIRED_CURV_WIDTH,))
     for k in ['lead_prob', 'lane_lines_prob', 'meta']:
       self.parse_binary_crossentropy(k, outs)
     self.parse_categorical_crossentropy('desire_state', outs, out_shape=(ModelConstants.DESIRE_PRED_WIDTH,))

selfdrive/modeld/runners/model_runner.py

@@ -0,0 +1,114 @@
+import os
+from openpilot.system.hardware import TICI
+
+#
+from tinygrad.tensor import Tensor, dtypes
+from openpilot.selfdrive.modeld.runners.tinygrad_helpers import qcom_tensor_from_opencl_address
+from openpilot.selfdrive.modeld.runners.ort_helpers import make_onnx_cpu_runner, ORT_TYPES_TO_NP_TYPES
+import pickle
+import numpy as np
+from pathlib import Path
+from abc import ABC, abstractmethod
+from openpilot.selfdrive.modeld.models.commonmodel_pyx import DrivingModelFrame, CLMem
+
+if TICI:
+  os.environ['QCOM'] = '1'
+
+SEND_RAW_PRED = os.getenv('SEND_RAW_PRED')
+MODEL_PATH = Path(__file__).parent / '../models/supercombo.onnx'
+MODEL_PKL_PATH = Path(__file__).parent / '../models/supercombo_tinygrad.pkl'
+METADATA_PATH = Path(__file__).parent / '../models/supercombo_metadata.pkl'
+
+
+class ModelRunner(ABC):
+  """Abstract base class for model runners that defines the interface for running ML models."""
+
+  def __init__(self):
+    """Initialize the model runner with paths to model and metadata files."""
+    with open(METADATA_PATH, 'rb') as f:
+      self.model_metadata = pickle.load(f)
+    self.input_shapes = self.model_metadata['input_shapes']
+    self.output_slices = self.model_metadata['output_slices']
+    self.inputs: dict = {}
+
+  @abstractmethod
+  def prepare_inputs(self, imgs_cl: dict[str, CLMem], numpy_inputs: dict[str, np.ndarray])-> dict:
+    """Prepare inputs for model inference."""
+
+  @abstractmethod
+  def run_model(self):
+    """Run model inference with prepared inputs."""
+
+  def slice_outputs(self, model_outputs: np.ndarray) -> dict:
+    """Slice model outputs according to metadata configuration."""
+    parsed_outputs = {k: model_outputs[np.newaxis, v] for k, v in self.output_slices.items()}
+    if SEND_RAW_PRED:
+      parsed_outputs['raw_pred'] = model_outputs.copy()
+    return parsed_outputs
+
+
+class TinygradRunner(ModelRunner):
+  """Tinygrad implementation of model runner for TICI hardware."""
+
+  def __init__(self, frames: dict[str, DrivingModelFrame] | None = None):
+    super().__init__()
+    # Load Tinygrad model
+    with open(MODEL_PKL_PATH, "rb") as f:
+      self.model_run = pickle.load(f)
+
+    self.input_to_dtype = {}
+    self.input_to_device = {}
+
+    for idx, name in enumerate(self.model_run.captured.expected_names):
+      self.input_to_dtype[name] = self.model_run.captured.expected_st_vars_dtype_device[idx][2]  # 2 is the dtype
+      self.input_to_device[name] = self.model_run.captured.expected_st_vars_dtype_device[idx][3]  # 3 is the device
+
+    assert TICI or frames is not None, "TinygradRunner requires frames for non-TICI hardware"
+    self.frames = frames
+    self.is_memory_model = None  # Use None to indicate that it hasn't been determined yet
+
+  def prepare_inputs(self, imgs_cl: dict[str, CLMem], numpy_inputs: dict[str, np.ndarray]) -> dict:
+    if self.is_memory_model is None:
+      self.is_memory_model = any(self.input_to_dtype[key] == dtypes.uint8 for key in imgs_cl)
+      print(f"Memory model: {self.is_memory_model}")
+
+    # Initialize image tensors if not already done
+    for key in imgs_cl:
+      if TICI and self.is_memory_model and key not in self.inputs:
+        self.inputs[key] = qcom_tensor_from_opencl_address(imgs_cl[key].mem_address, self.input_shapes[key], dtype=dtypes.uint8)
+      elif not TICI or not self.is_memory_model:
+        shape = self.frames[key].buffer_from_cl(imgs_cl[key]).reshape(self.input_shapes[key])
+        self.inputs[key] = Tensor(shape, device=self.input_to_device[key], dtype=self.input_to_dtype[key]).realize()
+
+    # Update numpy inputs
+    for key, value in numpy_inputs.items():
+      if key not in imgs_cl:
+        self.inputs[key] = Tensor(value, device=self.input_to_device[key], dtype=self.input_to_dtype[key]).realize()
+
+    return self.inputs
+
+  def run_model(self):
+    return self.model_run(**self.inputs).numpy().flatten()
+
+
+class ONNXRunner(ModelRunner):
+  """ONNX implementation of model runner for non-TICI hardware."""
+
+  def __init__(self, frames: dict[str, DrivingModelFrame]):
+    super().__init__()
+    self.runner = make_onnx_cpu_runner(MODEL_PATH)
+    self.frames = frames
+
+    self.input_to_nptype = {
+      model_input.name: ORT_TYPES_TO_NP_TYPES[model_input.type]
+      for model_input in self.runner.get_inputs()
+    }
+
+  def prepare_inputs(self, imgs_cl: dict[str, CLMem], numpy_inputs: dict[str, np.ndarray]) -> dict:
+    self.inputs = numpy_inputs.copy()
+    for key in imgs_cl:
+      self.inputs[key] = self.frames[key].buffer_from_cl(imgs_cl[key]).astype(self.input_to_nptype[key]).reshape(self.input_shapes[key])
+    return self.inputs
+
+  def run_model(self):
+    return self.runner.run(None, self.inputs)[0].flatten()

system/manager/process_config.py

@@ -75,7 +75,8 @@ def use_sunnylink_uploader_shim(started, params, CP: car.CarParams) -> bool:
 def is_snpe_model(started, params, CP: car.CarParams) -> bool:
   """Check if the active model runner is SNPE."""
   # TODO-SP: I want to do a little more optimization here to only check this once when we've transitioned from offroad to onroad.
-  return bool(get_active_model_runner(params) == custom.ModelManagerSP.Runner.snpe)
+  return False
+  # return bool(get_active_model_runner(params) == custom.ModelManagerSP.Runner.snpe)
 
 def is_stock_model(started, params, CP: car.CarParams) -> bool:
   """Check if the active model runner is stock."""