diff --git a/dataset_gen/src/grader/grader.py b/dataset_gen/src/grader/grader.py
new file mode 100644
index 0000000..715d250
--- /dev/null
+++ b/dataset_gen/src/grader/grader.py
@@ -0,0 +1,131 @@
+from random import random
+
+from grader_config import dist_grade, metric, pos_coeff, photo_coeff
+from task_config import time_limit, x_range, y_range, z_range, barriers
+from task_solution import positions, photos as made_photos, used_time
+
+
+class Point:  # stub
+    def __init__(self):
+        self.x = random()
+        self.y = random()
+        self.z = random()
+
+
+class SolutionReader:  # stub for reading from solution log file
+    def __init__(self, drone, task):
+        self.time = used_time
+        self.photos = made_photos
+        self.positions = positions
+
+
+class Grader:
+    def __init__(self, drones, task_id, grade_criteria):
+        self.drones = drones
+        self.task_id = task_id
+        self.grade_criteria = grade_criteria
+        self.metric = metric
+        self.pos_coeff = pos_coeff
+        self.photo_coeff = photo_coeff
+
+    def __get_distance(self, p1: Point, p2: Point):
+        x = p1.x - p2.x
+        y = p1.y - p2.y
+        z = p1.z - p2.z
+        if self.metric == "chessboard":
+            return max(abs(x), abs(y), abs(z))
+        elif self.metric == "taxicab":
+            return abs(x) + abs(y) + abs(z)
+        elif self.metric != "plane":
+            print("Unknown metric, proposed euclidean")  # raise?
+        return (x**2 + y**2 + z**2)**0.5
+
+    def __get_pos(self, drone):
+        return Point()  # stub
+
+    def __get_expected_pos(self, drone):
+        return Point()  # stub
+
+    def __get_photos_count(self, drone):
+        return int(random()*5)  # stub
+
+    def __get_expected_photos_count(self, drone):
+        return 5  # stub
+
+    def __grade_position(self):
+        total_grade = 0
+        for drone in self.drones:
+            pos = self.__get_pos(drone)
+            expected_pos = self.__get_expected_pos(drone)
+            dist = self.__get_distance(pos, expected_pos)
+            keys = list(filter(lambda key: dist <= key, dist_grade.keys()))
+            key = min(keys) if keys else None
+            grade = dist_grade[key] if key is not None else 0
+            total_grade += grade
+        return total_grade/len(self.drones)  # return avg
+
+    def __grade_photo(self):
+        total_grade = 0
+        for drone in self.drones:
+            count_photos = self.__get_photos_count(drone)
+            expected_photos = self.__get_expected_photos_count(drone)
+            grade = count_photos/expected_photos
+            total_grade += grade
+        return total_grade/len(self.drones)  # return avg
+
+    def grade(self):
+        used_pos_coeff = self.pos_coeff \
+            if "position" in self.grade_criteria else 0
+        used_photo_coeff = self.photo_coeff \
+            if "photo" in self.grade_criteria else 0
+        return (used_pos_coeff*self.__grade_position()
+                + used_photo_coeff*self.__grade_photo()) \
+            / (used_pos_coeff + used_photo_coeff)
+
+    def check_solution(self, drone, task):
+        solution = SolutionReader(drone, task)
+        if not self.__check_time(solution):
+            return False
+        if not self.__check_borders(solution):
+            return False
+        if not self.__check_collision(solution):
+            return False
+        return True
+
+    def __check_borders(self, solution):
+        positions = solution.positions
+        x = (min([i[0] for i in positions]), max([i[0] for i in positions]))
+        if x[0] < x_range[0] or x[1] > x_range[1]:
+            return False
+        y = (min([i[1] for i in positions]), max([i[1] for i in positions]))
+        if y[0] < y_range[0] or y[1] > y_range[1]:
+            return False
+        z = (min([i[2] for i in positions]), max([i[2] for i in positions]))
+        if z[0] < z_range[0] or z[1] > z_range[1]:
+            return False
+        return True
+
+    def __check_collision(self, solution):
+        for position in solution.positions:
+            for barrier in barriers:
+                check = []
+                for ind in range(3):
+                    check.append(barrier["borders"][ind][0] - 0.5 < position[ind] <
+                                 barrier["borders"][ind][1] + 0.5)  # drone size?
+                if not all(check):
+                    continue
+                equation = barrier["equation"]
+                if abs(equation[0]*position[0] +
+                       equation[1]*position[1] +
+                       equation[2]*position[2] - equation[3]) < 0.5:
+                    return False
+        return True
+
+    def __check_time(self, solution):
+        return solution.time <= time_limit
+
+
+for _ in range(3):
+    grader = Grader(['1', '2', '3'], 1, ["photo", "position"])
+    print(grader.grade())
+
diff --git a/dataset_gen/src/grader/grader_config.py b/dataset_gen/src/grader/grader_config.py
new file mode 100644
index 0000000..0a8d15b
--- /dev/null
+++ b/dataset_gen/src/grader/grader_config.py
@@ -0,0 +1,7 @@
+dist_grade = {1: 1,
+              5: 0.9,
+              10: 0.7,
+              20: 0.4}
+metric = "plane"  # options: plane, chessboard, taxicab
+pos_coeff = 1
+photo_coeff = 1
diff --git a/dataset_gen/src/grader/task_config.py b/dataset_gen/src/grader/task_config.py
new file mode 100644
index 0000000..f941a4d
--- /dev/null
+++ b/dataset_gen/src/grader/task_config.py
@@ -0,0 +1,13 @@
+time_limit = 1000
+
+x_range = (-100, 100)
+y_range = (-50, 50)
+z_range = (-10, 10)
+
+# proposed format for barriers
+barriers = [
+    {"equation": (1, 0, 0, 5),  # Ax + By + Cz = D
+     "borders": ((5, 5), (-12, 12), (15, 20))}
+]
+
+photos = [((0, 0, 0), (30, 60, 90))]
diff --git a/dataset_gen/src/grader/task_solution.py b/dataset_gen/src/grader/task_solution.py
new file mode 100644
index 0000000..1581b92
--- /dev/null
+++ b/dataset_gen/src/grader/task_solution.py
@@ -0,0 +1,5 @@
+positions = [(0, 0, 0), (1, 1, 1), (3, 1, 1), (5, 0, 0), (30, 0, 0)]
+
+photos = [((0, 0, 0), (30, 30, 30))]
+
+used_time = 980.7