updating clarity of the video

graphs/dataflow_robot_vlm.yml

@@ -16,6 +16,7 @@ nodes:
       inputs:
         tick: dora/timer/millis/750
         planning_control: planning/control
+        led: whisper/led
       outputs:
         - control_reply
         - position
@@ -33,6 +34,7 @@ nodes:
         audio: dora/timer/millis/1000
       outputs:
         - text
+        - led
 
   - id: llm
     operator:
@@ -48,7 +50,8 @@ nodes:
       python: ../operators/policy.py
       inputs: 
         init: llm/init
-        goal_reached: planning/goal_reached
+        reached_kitchen: planning/reached_kitchen
+        reached_living_room: planning/reached_living_room
       outputs:
         - go_to
         - reloaded
@@ -65,7 +68,8 @@ nodes:
           queue_size: 1
       outputs:
         - control
-        - goal_reached
+        - reached_kitchen
+        - reached_living_room
 
   
       

operators/llm_op.py

@@ -3,7 +3,7 @@ import pylcs
 import os
 import pyarrow as pa
 from transformers import AutoModelForCausalLM, AutoTokenizer
-
+import torch
 
 import re
 import time
@@ -142,6 +142,7 @@ class Operator:
         dora_event,
         send_output,
     ) -> DoraStatus:
+        global model, tokenizer
         if dora_event["type"] == "INPUT" and dora_event["id"] == "text":
             input = dora_event["value"][0].as_py()
             # Path to the current file
@@ -167,7 +168,14 @@ class Operator:
             print("response: ", output, flush=True)
             with open(path, "w") as file:
                 file.write(source_code)
-            time.sleep(8)
+            del model
+            del tokenizer
+            # model will still be on cache until its place is taken by other objects so also execute the below lines
+            import gc  # garbage collect library
+
+            gc.collect()
+            torch.cuda.empty_cache()
+            time.sleep(9)
             send_output("init", pa.array([]))
 
             ## Stopping to liberate GPU space
@@ -222,7 +230,7 @@ if __name__ == "__main__":
                 [
                     {
                         "path": path,
-                        "user_message": "Ask model if there is someone with a red shirt, if there is, say I'm bringing coffee, and go to the kitchen, if no one go home",
+                        "user_message": "go to the living room, ask the model if there is people, if there is, say i'm going to go get coffee for you, then go to the kitchen, when you reach the kitchen, check with the model if there is a person and say can i have a coffee please, then wait 10 sec and go back to the living room",
                     },
                 ]
             ),

operators/planning_op.py

@@ -2,7 +2,7 @@ import time
 import numpy as np
 import pyarrow as pa
 from dora import DoraStatus
-
+from constants import KITCHEN, LIVING_ROOM
 
 CAMERA_WIDTH = 960
 CAMERA_HEIGHT = 540
@@ -100,31 +100,47 @@ class Operator:
 
                 if len(dora_event["value"]) > 0:
                     self.waypoints = dora_event["value"].to_numpy().reshape((-1, 2))
+
             elif id == "position":
                 print("got position:", dora_event["value"], flush=True)
-                ## No bounding box yet
-                if self.waypoints is None:
-                    print("no waypoint", flush=True)
+                value = dora_event["value"].to_numpy()
+                [x, y, z] = value
+                self.position = [x, y, z]
+                if self.image is None:
+                    print("no image", flush=True)
                     return DoraStatus.CONTINUE
+                ## No bounding box yet
                 if self.completed == False:
                     print("not completed", flush=True)
                     return DoraStatus.CONTINUE
-                value = dora_event["value"].to_numpy()
-                [x, y, z] = value
-                self.position = [x, y, z]
 
+                if self.waypoints is None:
+                    print("no waypoint", flush=True)
+                    return DoraStatus.CONTINUE
+                # Set Waypoints to None if goal reached
                 # Remove waypoints if completed
-                if (
-                    len(self.waypoints) > 0
-                    and np.linalg.norm(self.waypoints[0] - [x, y]) < 0.1
+                elif (
+                    self.waypoints.shape[0] == 1
+                    and np.linalg.norm(self.waypoints[0] - np.array([x, y])) < 0.2
                 ):
-                    self.waypoints = self.waypoints[1:]
-                    print("removing waypoints", flush=True)
-                if len(self.waypoints) == 0:
                     print("goal reached", flush=True)
-                    send_output("goal_reached", pa.array(self.image.ravel()))
+                    goal = self.waypoints[0]
+                    if np.linalg.norm(KITCHEN[-1] - goal) < 0.2:
+                        send_output("reached_kitchen", pa.array(self.image.ravel()))
+                    elif np.linalg.norm(LIVING_ROOM[-1] - goal) < 0.2:
+                        send_output("reached_living_room", pa.array(self.image.ravel()))
+                    else:
+                        raise ValueError(
+                            "Could not find goal reached: ", goal, "pos:", self.position
+                        )
                     self.waypoints = None
                     return DoraStatus.CONTINUE
+                elif (
+                    self.waypoints.size > 0
+                    and np.linalg.norm(self.waypoints[0] - np.array([x, y])) < 0.1
+                ):
+                    self.waypoints = self.waypoints[1:]
+                    print("removing waypoints", flush=True)
 
                 z = np.deg2rad(z)
                 self.tf = np.array([[np.cos(z), -np.sin(z)], [np.sin(z), np.cos(z)]])
@@ -156,7 +172,7 @@ class Operator:
                     [
                         {
                             "action": "gimbal",
-                            "value": [0.0, goal_angle],
+                            "value": [10.0, goal_angle],
                             "count": self.count,
                         },
                         {
@@ -164,7 +180,7 @@ class Operator:
                                 self.waypoints[0][0] - x,
                                 self.waypoints[0][1] - y,
                                 0.0,  # -goal_angle,
-                                0.6,
+                                0.8,
                                 0.0,  # 50,
                             ],
                             "action": "control",

operators/plot.py

@@ -127,6 +127,15 @@ class Operator:
                 cv2.putText(
                     image, self.buffer, (20, 14 + 15 * 25), FONT, 0.5, (190, 250, 0), 2
                 )
+                cv2.putText(
+                    image,
+                    f"pos: {self.position}",
+                    (20, 20),
+                    FONT,
+                    0.5,
+                    (190, 250, 100),
+                    2,
+                )
 
                 i = 0
                 for text in self.submitted[::-1]:

operators/policy.py

@@ -10,6 +10,9 @@ KITCHEN = np.array([[0.0, -0.2], [-1.0, -0.3], [-2.0, -0.5]]).ravel()
 
 ## Policy Operator
 class Operator:
+    def __init__(self):
+        pass
+
     def speak(self, text: str):
         speak(text)
 
@@ -28,4 +31,5 @@ class Operator:
             elif id == "reached_kitchen":
                 image = event["value"].to_numpy().reshape((540, 960, 3))
                 pass
+
         return DoraStatus.CONTINUE

operators/robot.py

@@ -1,4 +1,4 @@
-from robomaster import robot
+from robomaster import robot, led
 from typing import Callable, Optional
 from dora import DoraStatus
 
@@ -26,6 +26,7 @@ class Operator:
         self.position = np.array([0.0, 0.0, 0.0])
         self.count = -1
         self.event = None
+        self.rgb = [0, 0, 0]
 
     def execute_backlog(self):
         if len(self.backlog) > 0:
@@ -74,5 +75,12 @@ class Operator:
                 if len(self.backlog) == 0:
                     self.backlog += command
                     self.execute_backlog()
-
+            elif dora_event["id"] == "led":
+                [r, g, b] = dora_event["value"].to_numpy()
+                rgb = [r, g, b]
+                if rgb != self.rgb:
+                    self.ep_robot.led.set_led(
+                        comp=led.COMP_ALL, r=r, g=g, b=b, effect=led.EFFECT_ON
+                    )
+                    self.rgb = rgb
         return DoraStatus.CONTINUE

operators/utils.py

@@ -82,7 +82,7 @@ def ask_vlm(image, instruction):
     inputs = {k: torch.tensor(v).to(DEVICE) for k, v in inputs.items()}
 
     generated_ids = model.generate(
-        **inputs, bad_words_ids=BAD_WORDS_IDS, max_new_tokens=10
+        **inputs, bad_words_ids=BAD_WORDS_IDS, max_new_tokens=50
     )
     generated_texts = PROCESSOR.batch_decode(generated_ids, skip_special_tokens=True)
 

operators/whisper_op.py

@@ -11,7 +11,7 @@ import sounddevice as sd
 model = whisper.load_model("base")
 
 SAMPLE_RATE = 16000
-MAX_DURATION = 15
+MAX_DURATION = 20
 
 
 class Operator:
@@ -24,12 +24,13 @@ class Operator:
         dora_event,
         send_output,
     ) -> DoraStatus:
+        global model
         if dora_event["type"] == "INPUT":
             ## Check for keyboard event
             with keyboard.Events() as events:
                 event = events.get(1.0)
                 if event is not None and event.key == Key.up:
-
+                    send_output("led", pa.array([0, 255, 0]))
                     ## Microphone
                     audio_data = sd.rec(
                         int(SAMPLE_RATE * MAX_DURATION),
@@ -47,4 +48,11 @@ class Operator:
                     send_output(
                         "text", pa.array([result["text"]]), dora_event["metadata"]
                     )
+                    send_output("led", pa.array([0, 0, 255]))
+                    del model
+
+                    import gc  # garbage collect library
+
+                    gc.collect()
+
         return DoraStatus.CONTINUE